diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/model.safetensors b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b5649bb1d94b29467512ead9d6332238322dab9d --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:055c156fa9247ea336c57677a75baf4dffa29770d4a87990db137c199b5d61fd +size 75550008 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0349dc38c1bd0c5e040e948be45fd7036f88375a --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a6a8e5821392a2673ad9443d80165d9971c1d783a679a4dd794225252951bc3 +size 99648 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..eafcf59725f9260f83d78e82ea9614cc05a19744 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8cc6989b6b5d7c9330003643df2d3a8d6c80fec40d7d49c290d71042bb12886 +size 50471 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7d637bddde6fbe7227861c6ef6bda734ce3dff6c --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de6696441ed391081a4675f1ec5190f5cff2611e67c40db501ee33700c86d322 +size 50382 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c01845a71bbe12825117736ebe608d5e68bd8999 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c2dee732e1daab82dac91e1ae554cd335fd00c983c186b2462f1f57c58290bf +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b6af57cbf8e8b9d3576e1c4c44e44932f8c71c4c --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a8ce5ad6d58c2268628d0cd710e8b89eaf34c828b724f6cb73809daa7b4fe07 +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/model.safetensors b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2e3a9aae10f93f4ef9f1b215c91683475fd773a6 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:baefce8eae7482a2fba364fb1ad24a75c4f703ce7ed0be1acc650836449a77d6 +size 75550008 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5f90325b2e8f10c82d9ec8745a42cf4ea9d92aac --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:618e7abfa3fefe45e1729cf22b3be484fb4697b4733d8dc40e1ce8bd48ed6844 +size 99648 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..c84af5c0082b407a4061f734eabb4f6dff39e57e --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69402c34c483168d4266fdaa50d5dbd21a99c5ad515eefb9d435717735845e9c +size 50471 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0bfd72a8625ace401067893a4c69ff78399f6dd4 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5d7e43e77532066a9db575c049ef2c5cb391fd0bec038e0a1340fab458594463 +size 50382 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c4ce2c8eafea141db3cc4db115ce9f1cc13849e6 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:672e5f3d050f699c9ba282bcbd398c88932fb419bf53f852d33f3f497fe7569f +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0e9c1a8d10c940a91f7cf1f50beabd79bb4d45b0 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7079387f30808005335f02c85f8167a8dab8db798d3542975b2ee4e2b0f2e75 +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/model.safetensors b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..17fd9e1a29ab78e4c79aa199ceb34beaf5c793b6 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e82251cd781aba7d2e8ccd6c851740f6a7c93b20625856be8dfdbbfce4cb9df +size 75550008 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7ef7f3ff3f49ac7d5b436911f5bd21385e6fff90 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6cc1981c84187753a00b48a66c73281b49e1ad29dabb1346cc7631b245efbc9b +size 99648 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..dc5449381457ac5579455e7d2861f607179c6a60 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:124c54528a55d170391bee0124e5d27cbc114e32340e312cc616cff9502459ad +size 50471 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..6389eb5f41dfe9852f767f5cb97942aaef4c346e --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:798d46a417dce0a07eff3b5276bd15a6e650efb7683ef5de0e62334f937279d6 +size 50382 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3558ef3184100a27493bf50e49bfecbb59612b9b --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff590d3dda8191dd0d36d44fb23024f9271c14011379987412647344adab0412 +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8e1c61badb1a65d7f7b4d055021d84c625d5e96e --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b0fab2fb291289c5ba1cdb09a2905e3ee36f36babd18a1148f6983669c0c1ed +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/model.safetensors b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b954fabbb823a2732e19a07e029dcac6329b354f --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0242761dd49746eecd50377415fd0843876b1a9e874e253802011771411df463 +size 75550008 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..be3820dd40353b24e26096e82d91e11325a13a21 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c43f9d136938e92a8d2cc2dab7a5fd244f7601a1f54db06ca9b915b2f1dd520 +size 99648 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..9cee1280c8e64152e1c3ff3a296cbb508b7ddbea --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58591aa3261907f94371862b33d8d0ba3bc8717d6ca9d74513fd795ad23f9ed0 +size 50471 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c95c7d4b62c1f4c74712311f1e0a1e2c7e401441 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d15d561a855d96b74e3ec87af7582753a05f7856dee11ba64dfa6dd5597c7ad +size 50382 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..187abb84c03c6f5e4cdbb96929cb3bca417de84f --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b97971c84b75d93d936c52b0c2ca38fa06e6bc3a9e14f115126d2f82ed9d471 +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..733c8024afd179ab5ddc03f6230657826528bb4a --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d05a27835087adc265e33aac043ff9b6dd0bdf45fa56435c3eb3d13758020cba +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/model.safetensors b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9afd849405b13182aa06c9b0211a3b69aefe245c --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:104252530b3ef1227e6c127825a99feac5bbb7e6075f965f46552df13b59e017 +size 75550008 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..db8a0382636e9e762af7adf96f77f4c2909f3229 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:086bdf4d141d538110e53c6506b1fb894bd8c7ef7178a72e9d265a6c3e50258b +size 99648 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..e52c5e01e10eae8b0399d9155d862944b829d092 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:503add349e5817f0241720b94503441e91ef365ba3043055e5b00fba6ce98b5e +size 50471 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..521c1dacc531a5c7f28ed9cdd43dcb6b97a4cd0f --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af5770ccaa3d389acb40c1e9447d2407c9333f0d0ad461f1fe270f1f42ff54f8 +size 50382 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4914555bdf963c1c1c0606e7ab7fb3a1aa60b60b --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82d1abd0cfe5435090b47a325fbc2bbd666e0bde129bba1336bf08ec64a59400 +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ca1d2a6922c7a5243d4e1c8f7b1e7e2f272ad37d --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:064414d206d7c65e568169df55a735e2329910a5f8f21808635367f1743e6332 +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/model.safetensors b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9700fef694e616b6f5f804e9c8852aad028bc296 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:572f873615b01bbc97c83b8bc6093a794db0e7c84950f86e599c58fb2eb00a4c +size 75550008 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4ab8aeaceaf3872333db6cbe10997fc087814d1a --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3142fc300d3694ef193768ddfd619b31215197c1e2ca7e11ecc988c059e4237d +size 99648 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..2248d35f28e46830b9b69aa9a1751b9bf22f7ca4 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec718c72d26f376af1cf4c0185e89fcbede80cfa7c324372e07916ad33ec836a +size 50471 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..57dca52a3de28e2b47a92976f0fd0ec8b0a3a36b --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2de1ee087b83632b8e2a0357f561e4750e12229935135a9b5220cd5092e1717e +size 50382 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..049e99f9c43ab1fe9cf9447fa2ac735d4d32e7bc --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12d372b853de9540f39becd3aad27dadd392e3d6048969a4abe56099366165b5 +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f7fd5f62cee0ae726516d3acb0bac25c6f05d257 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3288ccb3684d92953422784de1aa623a848a73801f1021e75ad325a0d774280b +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/model.safetensors b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d31cd754fe99330e36d1f5fc0420388e3941c6b4 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c8a848e9a23df1f020cc76ce72bd526ff0ec074e437418affb8f437aa9288df +size 75550008 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ebcbef159ff06e660cf86b7a37d021f95fcfbe2 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3588006609dd506a8186ccaf17d37deec201389e5a34e6da62289dcc94eca79b +size 99648 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..9bc2b6e80067202955ee956b3c2c2c22628f8704 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a9781d102fb0daee6b144d85c3da987f15f0e6daddfe1d0f7d4812acbfafbc7c +size 50471 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..08f0dafe72fc25e831b92b6ad4a1a04dc4bde6df --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1463d1b360a4beded7970c985d35f791a60775c3da31dd108440c1d86a8b7911 +size 50382 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..80f8a1925fa34dd1d80123d11f2b68692cd07fa2 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11241e7cda520635f3cced0664766216de3b34c53a87c90fb9cc249bb6815dcd +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c823dc0e791716a4cd4942a44a0362d1a5a290ad --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08ae513238ba5dee5560a729b2d631055d3ebdefc6c70898b550e4a012a6ef4d +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/model.safetensors b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..607cc5fccd3ef3b96b42a034ac1ed7b7edc4ce10 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74b61abd662c180495556625a5787087724871febf771dff7554c73d16b798fa +size 75550008 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0f5f74f11a2b3c86525c92f0f3b2770ee09bbe65 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3cb197d4822cdd32ac5dae7b481036264212c044409b50215744b2c690af516b +size 99648 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..6cd4d990a4a0447b0395ff58e509f75f7e1149fd --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f54d9dbfc687c397f8541581b1022199fbf7e7a17f97d80298ee90f1a0977efe +size 50471 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c34869b8fc48b21f2a7d22c0902db1501ddbab7c --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f36fad01262d9e53d5aa3855317fe9232dbe9a3801084f4d8a35967ef068d84 +size 50382 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2faa39b32123ab5810751ced9354f6f489602725 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d600f2ccf9393c19956c02d98efd5b014efbe29ba565813ba54d0f8dbf4c937 +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8400da471b645bf77ada02024142d79c3294a0d6 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69eb53d9b823c3ce7d8269b422aecc39084b220bd1bbf278cb3de34eff9be9a1 +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/model.safetensors b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..af0fc9c1b384fe8d7a5277710a790c3a43115e55 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de9ea9b3458d2fb299cfb6d6138c2993656a756abddce2a882eae3946fb4d10a +size 75550008 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4cb84f2f7216e0417ece2709a69cdd78b6729019 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f39b70da66c5a23138f3ad387e937392f517ec5843f79e9405899b4b4c74029 +size 99648 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..64d0de3103a996d5f4cf15d6a845c856edff99ab --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a99a740f5d0c8b8e3430f9053d1e2d2c59128400bd878e0e9a505025c2b9dff2 +size 50471 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..76d08032390e19a2256584b4f8d920d6229c8a5f --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b330d5ed1ee5d97ee382db06ee585bb7a278a8937adf35e9cbca07c7bdb50fd7 +size 50382 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3af717d68b68136f5604d943c20dc091e87172a3 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f38d485441cbc3fc6460e1f74dd506a34cec44b3daa7c8c63e67233f94c5fccb +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2a59fd4b49d9dd184e07f79352ff554ad679a1b6 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:15d65bc88768d8b225151e68436a3ace98fb06b4b46b894290b7d9e1b92cfe0d +size 50387 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/model.safetensors b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f1e771b3b2fe836139d0445aed614b5507ed8e5a --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e403bf43b9ccca3ee09be0443edc405c0df6d3b4cea2f01e69a99069798e3529 +size 18890032 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ab103779718cc133aaf7dd2f53fb4598b978ed87 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8bf5fe4405f05ffcf9867f9b64afdf3fc170b26e8134bd97f0296c4f4609ec73 +size 25920 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..702ce956bc3976b7312f3363b4d6553ad1de50f6 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c905701bdadbf9f7629adb236bd672cfa46a6b8b96c65ad86ba5382846a127d9 +size 13607 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..35b1b89f81df19c887e20a8025efcb8c6fc62af8 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2dc70d8d059e4c0a04cd9037738e2f97b4f4db2c18476c738747b8ced18990cd +size 13518 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2dd5ffd34d0048548c7736fb55f2fbc465ea1250 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f12beaf28920336b342e68b33fffd2f97145bff18497c436f16b3d71005d5092 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..326c66095bf708cea2ed15a1ed0fbdbb946ec5ec --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd045565ecfc2984bde109c5c3e424e1627f1279592cad89009b4a4026a2d25c +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/model.safetensors b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..982f16f932daee51770a91c0ad52a36dd358e3fa --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d574ba8dc4303c72573a41b41ef096785b211a0378748f9152154ddc9fb7f8f +size 18890032 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d6ef94eefbfedfe05002f4093cb1e067240e5ca9 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:597f4ce5cc79c18ecb1d691a05aefdde604ba582457cbd5b2aacb2b03c942ee0 +size 25920 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..f2cb8373c3664a48de7f833d39d0827b49de0475 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db16b787e2ceaa9a12c899b3de04acab73e9289436603b5a9f55a10f162cd141 +size 13607 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..73e19e493496b988081229a1bd0f840591ebb733 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66727442b48d23328ca54c20ecccb06c5bd5e43c1039fa61beb9d8bc3391f7bf +size 13518 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..977171e55ea4a498ebe4fb2cbcd0a3241e996f24 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e9f7b059f53441461504130a627d3b7d6260c2ab309460e9c1a9c7a0657290f +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..fc3e1fb37245f72cb117187c693948fe90714232 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5414b2483de646883bb4b38b586c0e495a11f152696637b569de13c4789610ee +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/model.safetensors b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6897e23aba98343956e361f52f2be56e1ef1c9a0 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d014f69c843b985cb30418236e5a18190448f0dcc51f5dd1c68db81f0d06989 +size 18890032 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..6a2946826e01ed607a412eb7e20b9dda0580a106 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9176fe63c8a678f8ba281ca581bad547f52b0d2a45bcf9daa92d2a0b9d01ac97 +size 25920 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..89c552076ca6d4eb9afab0e74776b2704152f066 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:839daf0a304d55f31450b575876f80f87054fa5782a6fac7b260c7e323008d20 +size 13607 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b632c58a7f136c01c6f97faa7871623519fc8183 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74894308f3e64abcef71d167ebb936c6a584e137cb8344006905b5ebae66ea5d +size 13518 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3add25d475d7dbcf90a21a1260dee0b4bbbedc07 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c24cc07dd171149e747cc92426f9be5f99c26d11413ffdbef036ef7533d6a8c4 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c31dfdf18f4eebe657460e1ff02a7c5677961ed --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7523719e38cb92275bb574f72b1b5bf89824400152b0067b83ae72ae0b0bd36f +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/model.safetensors b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3d400485c314f035780b07e0d1feee9a4264a276 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52723e0a950565bb160b1d589eecabd9d4debcc9d2cd4867fb470fb224699feb +size 18890032 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b1a4e084b5335a76d32ed3bf6f3f4c08205e1d99 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44f48544ec9bbd6f93d863bd6cc8eb3f94077caace1c3a91ea72e89d07287c6f +size 25920 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..4346fbc76123b46c6f950e830433334c3113caeb --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb8fad2c68b120be2e714f4e365ba047d77c73eac6b7837c6ffce03e6d3d32cb +size 13607 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bac59f5061c8ca491052f8a6dc136f84bc410957 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f64953a39bb009180117c3d8de648ceb12f349d28210931d357f66ffc3e9166 +size 13518 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b5ebd7751e60198547ac431153784d6f4f7c6fec --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01b8ca7938800d98f2f5a67bd89a41006e6864b2f452146f2216786b41bfd486 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..46214dbf9d091b05b7c99ce8b38261550ca97b65 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ef7beebcec98a9284ee4f88985cc96bfcf79b8a13b859f6aa544bb320606440 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/model.safetensors b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ad8189df5694c76fa82af455241fbfd17bd6b04d --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5badfafae8ba1163169c9185b32ddbcb3f9a69799de38487dc3e3500c2480d39 +size 18890032 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..54dbe44d191492c138cdfe09b9946a429ba9f67f --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0a23b042492c7bf8311b2d64e0ea5543b6fc393ffc3de26dc679f58357a99bc +size 25920 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..69ab2f79daecc050635131c6198ef5b95b064be7 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11e23096930a0ce30884936a7dc631e3ea927fbd69f39b36c56fe98d1b1dea9a +size 13607 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d8ffeddee178aee09ce67c1ebeaa047ceee4221c --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c33596933da46752135d275d816a1490d26f7969f7113736b378adf9298a76bc +size 13518 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..592f21fa40e4085cf05f2f7cb5f3340c96b10f31 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78d557e1d7ccc47d855e6e13759508716b52533ab42940c7ec4f1127555f2e30 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7db208fd579317cd24aa1e66726b39c0981f5550 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6c5785adad0025dceb5c39c316820edbec1e942c51ddec2f09b9797e4c70304 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/model.safetensors b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..be1ddb03e175ec348547fcd4b6e4358882bdf8e9 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab4787b3efe3944c026020d388179597c2cdbb21c6105594b99cab65ab8ec32e +size 18890032 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e72e500057afede2c19b3117ef946f3d2b517bf8 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5077f96dc67b8560cf26a7ec10482a18f1f9046c6aa4f838580e8ddf994a4f38 +size 25920 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..a6472a4a4c71bf687c0b6d3d77e3bad7cef0a1c2 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2009ca50a807b0c88e2684ac43147d747fee33c2855c369dc6716f82a169468c +size 13607 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..1174f5865b4080918ecd8446214e1556f8831cdf --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc0457ad3316174b39567a4211ff211c884f23d3510e9cb9e74c4abfb2fd22a9 +size 13518 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..54955ca8649a8dde4c3ee608040f97e01ed8dcbf --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd6e4cd4fe7c5ffa03aee455af9fad48a053ac4e91231c600197d45b39618b57 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a7452c1b0bfab882701741b910cc258e26a95bc0 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe8ad7b5b796be0f2409940b592f99b176af6fe1decac3663d415e521cd080c4 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/model.safetensors b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4d323546aca9ee096d38ea14e677a1d99e85dcc2 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:81f4684110c834d496c6ffeae20c29b761d407d6e573781fc0368ffc28aae45a +size 18890032 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c806b7cb3d5a4b401fa125c9d24a8317378f743 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76533bd103c704764bfab7889496df260c76bc3c74afa12e37c2d3465406be31 +size 25920 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..188acca95301ca2b931b9a7ed878825bd922f4f3 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2093ebc455ffbdd90c6a867c7194cb9a0e0c428dfae5106d33f2daab33b080a2 +size 13607 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..330e7f93e28e86d0a24fc9415afd78f922965646 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b16e64fb0e7a4f24db6f29177a31e85336d4ddf2bb3de7580b93cbf141d45d42 +size 13518 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..84d4e8240715d7696f1ae9272e90786615436d77 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:964d8aaf2d786300538129a6404ba3f23773f89d4094fc83ce5ad3b8fcbef179 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4d2611ea2fc42ccd80415f2f8a09dff3011498fa --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:607cbf27f96ad4a6499334ff3c2abc254cdb3ba029d9592c626825e952afccdd +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/model.safetensors b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..855f96e42803289b6061e7bea542c52252e9ed97 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7e80a20fb993977434605dddd0927d1fa365b647196548828a00e7d3b4556b42 +size 18890032 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e337553a1dfbc295028cb94e24c805c7e3874d0d --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be45615f0a617c49e5020e8816576e446d6309b3845aed599d3205ddc6ed6777 +size 25920 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..0591631c7b6115f146850273c9b63453cb27905f --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1c2c21d6e7ec61c1269aa0c52b0ff873d845a37968e97146304cbab91083e3e +size 13607 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2b757139198a97248f9fa0071fbcfafda30d84a1 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2628d52a5a7806b85e994c4acde37d148c22d41332b79a71c7750062883cfbe +size 13518 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ff2e0392ef69c6cf901f02ceaa0a678d27d32148 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:37a7c4aa82b9ece0adc32b7f02ae376c11295c41b4e9d3034374eb78d496be61 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7a092aa5a2c8118ada52358845a3bf952cff1a0d --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee2cfcfeec621526f51ca69a42376c8e2f4cff512745d0fe9299d577884c58fa +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/model.safetensors b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4fb78969b2337608e11deb69d7599091bdf0b9b4 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:093acaa06cce69536e79c5ae9b3f2922d8ee123b7cafdac967c3a3309e99ef49 +size 18890032 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..22463d093abf245a518ccaeaf3b45642522577f5 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:959646991bec68d08e425a9a86da7f7f0e89d498afa01bb9871c47d0b353361b +size 25920 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..934bd10f275da1cd68256f29237f23dec0610e20 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a5a432911d6eb5128360933bd374fccef592c0a5ff6d683daffe71230835149 +size 13607 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..60f279cedc089fe8258295c9d4ef8915ca8ce55f --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cc6b0c1929b010adb8b9aef16b14301f9ea21b3bccb075cc2d015d459d532a66 +size 13518 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..af1b8e330daaee9f1dbb91615f3b39ddab333263 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96e8b5fcc738b13377abe2c26f9b1264bf75148a5c5425e82018a7cb58389f71 +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c6c0dd1ed4dcb9263cf1f256ba4b18922420214c --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ff6ae3b5366c8b9fb8caba041870c4e63abddbb4e9334775a9530199c59d80f +size 13523 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/model.safetensors b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..27202e870c984fcf3d73622035a0176d14c9d6f5 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0211bd45de9a624119f75e0cf6f7ff554795002b1f4fdc9066f80482ddcb63e8 +size 37776696 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b9fd9e005758d43f5752c0add94123be1696908 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:178f44257ec4414be892c7c77105ae8ba4ea11a2e1ef99174956df8940b2d922 +size 50496 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..f625ebfdaa3bb893f440cb86fa3230c56bd1a419 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:771d7e0af4449e0b25201688fd61a1e7d41dfbec2b00ebf00df155a96d9217db +size 25895 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8d67688e57707601582d40e93486cf48fe059fae --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d5b6e7782eb3d5bff9a66208593a9844b4d3b3584a2f5ef95b76439a08870ffe +size 25806 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b51c682b1b041af4b2863c7e6bd789c01a341d0d --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29a4a95df44d9196bffcf48050c30182c1f67b6335588aa81eb6e4c3ab88bb68 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..03d12894aec5194db83b78175c3c3230a35905b8 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0889984c42573f91badd736549e70d17246b37d4fff1af44cd2eb2eaf497292f +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/model.safetensors b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..41f5fe674ef1eba517310e3ca265b1a4916770a2 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bef155c3eb0ef9d58c403628dce55ecf9a6a3aff2a8cb9dbe49fe0f9ea92ae25 +size 37776696 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..db2453b4a0ff66c8dbdb0bd2bf27d14e440f2ff0 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c4fccd2294d00202e6c0cd940bb372079da4b0ed0c18ca8947cc5b35ef31a5d +size 50496 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ecf05ebc40b31e3b643720c9acd24644e645cbb --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e1506f816632a37e37a356440daef6c222606ee558bfcccf0ab452c5f0e0a40 +size 25895 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..1272212c8d84c5fc46286ac04fb9045e4262f6ce --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a791f2b02113309e4d6d3c111d0284fd95c6e5113fb3401f510c04c25441366e +size 25806 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..02dd6d955da235e3a761aefef05dda993155040b --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5409966e3084d35c3406a580a7c9b80f2098ed214dba30a8c1d8ddc85b1ba25d +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4c59478fc47760bf97ef191b10bfb8dc5fc4f248 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a4a6a678216c5903acdbedc16bc5bf2f2c64a7f53e44f9ad57d778ba6b71f80 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/model.safetensors b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8fc11f66c3f0c3fdfc1366f10041c03d700dfd9e --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9d30d228ec398bf5986822c8767d657abe57b23a61f7f7b9fb8ca23f8f2616d0 +size 37776696 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a34a690d4ad665db4409652734e37bb6d6db7ff0 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6282e6db8ba363464d53323220c94663db76d6db738b602b6d3609ff2c31441a +size 50496 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..271c035a836789ded2de792fa227f3dc8cd7d340 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8efaa0ce7534349e129c0fdf46198ae333a3b90f6421c5d60aba4b62fecd5b66 +size 25895 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..760add5546c92acaceb311d1fb2feb9e43deb666 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbd4c297668ddf38f14316db7c5701680f417b98f9027d1170f357f34cddc1b6 +size 25806 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5271b80f6a351be344acaddc7f3cb33286860949 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc44d1c671751accd1b587ea2776d9368dea317b14cb5e70b11034b2ee9285a8 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a0e6f573c710cb05edccd5b5ff0636c1c5cae9a5 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:afbb5eeb0c5317bd4aa9b2111e2071d5e3a406b841331cbb8e1d851e23b80ccc +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/model.safetensors b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..25fc8dd2f37358bf615769ca12302b07e8efd890 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed60be3d656226647cb38cf7c132ba74165c3f4225985862d33e70bfba9c9eb6 +size 37776696 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4dac03408e526691e92d1344b6c0abd64676f2e1 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b8a4cfd7eaf684911ddd3bfdb5b50a64f76af6627b33663cd13e9126d791fe3 +size 50496 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..0df4edfae7cb22d4fe8a1378d86ca3844a213207 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:678c160668635d4fd0677ff59ba1e98c5d2c13ff57d37758b994253b8d173af7 +size 25895 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bb848e8efcbe54b80427ab082cf3f54aee12e428 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7eb1c4da1c68f6086342c576a6406ab459c005ef49554b9a988a673af8b5980 +size 25806 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5dc6ab0b68e11ec3e7242afdf1b85c1e309d7703 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d37463a1db13337dbeb11ced6d6d9c7dc045a8178ec91dcadee56b51237b53c +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3ce145198e16132759362099cc9d874d558ae243 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:15f6172d4e5d049eba524c438ac4279a18799010c70276230b1b4beae1fbd8ed +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/model.safetensors b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fbcb8bf374947804e137c8f9eb270e9bb712747a --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab7665c4095b82f98450480e3d611ae146dad858279e0d63c3bd472c9e6419c7 +size 37776696 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5b6fbee38957eb84935522ac8551d8367a6d3156 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02479371458af5a9c83317e561bbc5c0022a2d094e3da84b99e71a063f6f215d +size 50496 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..39bbdf2f01056cffdee94ba2d3a9b47eaac510d2 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a6a26647bef353201fe92198116128847ccd25e863bb7ce9508b684af46e34e +size 25895 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ed00fbca9c02d8bf91f55e607969c885b7d650a4 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7306290ccc58f4d6fd537bd45626aa2f09053eef338db1e4ee315107a1dbf4e +size 25806 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..544bb15525b5894751e2ac7240fe0bb52fbbf939 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0285052713616d8edb999ed917471219f0c5c5184149704f3e910cacde4ea1a7 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7a269d746630049cbfa54b4098252f9b3357aae8 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f1b96e610515f8adae3d14e5fd63483c0bb47e47936022db1bf1ddad2d89764 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/model.safetensors b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8f4886dd572b004741aa1a4664f1fea6e3d7c0c6 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c8ea33284d42f70364a0d3553605a13df12aa9b6db7007249cc3da239139b9e +size 37776696 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c458f0a8e5b348c16ac618edf31f70485c49b774 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b29f02760a141d7660c36bd3079297d601ab63f393055681da2e6891e38bfdb6 +size 50496 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..4e4700e41981b5fb26ad7e29747a0fcd2f7b28a1 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:717dd0faf57a8edd98035cf00af26035dba7f6a10c706b88df3e9ceb8d2e4ae2 +size 25895 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2dd98eadede11798f02b3b335f6eced880a09ba8 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:393f002c5b74a7744c0619b1a94f0ffd4a231917dd8779797859bcbd21b092ea +size 25806 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2dd51049969c0910b2133ac0f652184b4ba21e11 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50795e0b3e183218c2e3925a691e2eb3df0107ef358f880a907af2949e823ad3 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..88536674694a4a2d969f92c77be01dab51bad0df --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:878a5d4370eebccb8588845739972e138f0551af0a8d1cea45edd93eff35f0f0 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/model.safetensors b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0c54554216076ff275649d8317c47cadc2f1a4f6 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ead7990a402621c06efd1e104374aea91b388ec7683e01f8ef2fc8693fbfbe7 +size 37776696 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..06a9358688225189c09881002769065b31d9c8cb --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3bd2dde561943e4dbb527ce8017ec71b4ae3f4b7eb6c7bed980102b1e8eec8bb +size 50496 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..6041b370e86a71327fcc37cd8b1ed8dacf36e5c0 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd29406e87ae9478bab4133a1569306b30867bb0c03cec6f13007e6068616db7 +size 25895 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a327d6d6c69635dec2fae60e845e9c70684e8640 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1054145360c6f0c57f668b9cbd54cb432c2b9a2b8fe23fa9032a95ab1d8ebf5e +size 25806 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c6b45e974d88af9dbde04abe915d15492ef8cfef --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8680899d356b96fa389c3efe7c020daf493378c127f0bf0908b6fc404d3498a +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e7f1be5372609cd22f90efdafb6092bbacc6fb05 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a6075381dcc1def29a48c4ebb6e139085d08136149a97b791aba9952e339143 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/model.safetensors b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5aa821c09536d3f5763df39a31a06fc0d8b37440 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c5855e8465ad843f84ea64ec598c6a82395d6ff729da8baefc396dfb002788b +size 37776696 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4d320a92727ffa78a87d60cc8e6095466091ea55 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de717dd1480e8fd964b80123dbdeb64c092ac21a5c04a4b7d59fc4d6f792b2cb +size 50496 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..37e67009fd2081dff4364c0581219321f5856701 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a51c86f87519a448b495a4ce9f1a876419bed97f78a481688de079922eeca803 +size 25895 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f1d3695950b190b5a615fffc95550a234f455787 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd63908f1204fb3a6c409cabda69e90036eb7c52538804e9bcfe8b28fb7d7463 +size 25806 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3dac141e0f4de788ebced04bf512345cfaf98b41 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a3eceeb52816d69017f2ebc3f9a3f0430b887d35eae72ff94da88857133e863f +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..12d05d493e984d914d8dfed4fa4638eba8ca8a3a --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01e4f8e3e96cdd7f71faf0bd5de43e5cacd715b9ef6edab88fdd7f65603acfa2 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/model.safetensors b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..80c07daf4fc6bb279d1313c71e6dba112badfbd5 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd8ecaac6943e151ff9f1c584a462da9500472ae481cb57b3b48fbb9b0d1d513 +size 37776696 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b849b01ba9b33282e0de0b01de079d42ad215423 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:789d93778e7782ce3f632dc9f8dbc39161a0ebcfd890e8ce26793257e60ae12c +size 50496 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..9e61b1313eb7cc1ff42a6cea65541f9e48220948 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a0b267cea43c72aa92990fe8d9a7bdf71dcae76cdbd9084a5cb7b03da71e780 +size 25895 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..cd8f407e79f99aa453fa0cfe4fca6c1acc4da467 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1fb338a10a124488a5de02066eaeaf8735fdfe0155445adf8087018e84dd6c4b +size 25806 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8fd16749567c4e78fab78770f2b20d885fb33f42 --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a10d47f17ef245fdae1929b75880601f15d96ee518eade1eb8121381282b4820 +size 25811 diff --git a/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f7830215c4f8cdb662b9ff3de5db4a66a74935d --- /dev/null +++ b/lm/pythia-160m_layer8/consdl/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:07b536bb6c5f9edc0e5c35d43f090a8cb0e9527eb88489841349bd02739f61cc +size 25811 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/code_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x16/l0_128/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..5c8e9524b1bf2e076cc4180bb37671384fe22f12 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 128.0, + "recon_r2": 0.9470236410499802, + "l1_per_token": 4.743606038069484 + }, + "seed1": { + "l0": 128.0, + "recon_r2": 0.9471136835477139, + "l1_per_token": 4.732008654381236 + }, + "seed2": { + "l0": 128.0, + "recon_r2": 0.9470559093714713, + "l1_per_token": 4.741950588465082 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.4182974829931935, + "mcc_abs": 0.4260022991242523, + "act_iou": 0.2064467360584473, + "act_iou_n_pairs": 12288, + "input_r2": 0.9672127523129951, + "matched_mcc_signed": { + "0.6": 0.7677285457374272, + "0.7": 0.8006025489924771, + "0.8": 0.8352966298668986, + "0.9": 0.8929508958728128 + }, + "matched_mcc_abs": { + "0.6": 0.7725201425786331, + "0.7": 0.8011100285285565, + "0.8": 0.8335159045760728, + "0.9": 0.8913016852847915 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.42363765269119674, + "mcc_abs": 0.43109993172981786, + "act_iou": 0.20818798111939704, + "act_iou_n_pairs": 12288, + "input_r2": 0.9673380628863562, + "matched_mcc_signed": { + "0.6": 0.769089677063175, + "0.7": 0.7996713626757554, + "0.8": 0.8335694300763983, + "0.9": 0.8924420288784642 + }, + "matched_mcc_abs": { + "0.6": 0.7717117645072608, + "0.7": 0.8002973996121837, + "0.8": 0.8346499509540946, + "0.9": 0.8927197536566601 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.4171759314904439, + "mcc_abs": 0.42415803222992143, + "act_iou": 0.20668773024271417, + "act_iou_n_pairs": 12288, + "input_r2": 0.968432516148108, + "matched_mcc_signed": { + "0.6": 0.771314545473405, + "0.7": 0.8019411266983052, + "0.8": 0.8368953875817072, + "0.9": 0.8941887207725391 + }, + "matched_mcc_abs": { + "0.6": 0.7741185124516646, + "0.7": 0.8016780702599143, + "0.8": 0.8353089182716223, + "0.9": 0.8909427521109946 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 128.0, + "l0_std": 0.0, + "recon_r2_mean": 0.9470644113230552, + "recon_r2_std": 4.5619351541649955e-05, + "l1_per_token_mean": 4.739188426971934, + "l1_per_token_std": 0.006272717049954086, + "input_r2_mean": 0.9676611104491532, + "input_r2_std": 0.0006709886356966943, + "mcc_signed_mean": 0.4197036890582781, + "mcc_signed_std": 0.0034527556881677945, + "mcc_abs_mean": 0.42708675436133053, + "mcc_abs_std": 0.0035957648056547556, + "matched_mcc_signed_per_threshold": { + "0.6": 0.769377589424669, + "0.7": 0.8007383461221792, + "0.8": 0.8352538158416681, + "0.9": 0.8931938818412721 + }, + "matched_mcc_signed_band": 0.8246409083074472, + "matched_mcc_abs_per_threshold": { + "0.6": 0.7727834731791862, + "0.7": 0.8010284994668848, + "0.8": 0.8344915912672631, + "0.9": 0.8916547303508154 + }, + "matched_mcc_abs_band": 0.8249895735660373, + "act_iou_mean": 0.2071074824735195, + "act_iou_std": 0.0009434657079194839, + "act_iou_n_pairs": 12288.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/dict_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x16/l0_128/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..09018e78ec67c602d408a48ed51612fda68a881f --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/dict_metrics_step_final.json @@ -0,0 +1,772 @@ +{ + "schema": 1, + "step": "final", + "thresholds": [ + 0.1, + 0.2, + 0.3, + 0.4, + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 12288, + "d_in": 768, + "l0": 128, + "recon_r2": 0.9472764396198932, + "l1_per_token": 4.737958070614665, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dup_signed": { + "0.1": 12172, + "0.2": 361, + "0.3": 51, + "0.4": 11, + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.1": 1, + "0.2": 113, + "0.3": 11, + "0.4": 2, + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.1": 12172, + "0.2": 54, + "0.3": 21, + "0.4": 7, + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.15274138748645782, + "n_dup_abs": { + "0.1": 12185, + "0.2": 594, + "0.3": 72, + "0.4": 16, + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.1": 1, + "0.2": 183, + "0.3": 15, + "0.4": 2, + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.1": 12185, + "0.2": 126, + "0.3": 39, + "0.4": 14, + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.16024206578731537, + "n_dense": 240, + "dense_rate": 0.01953125, + "dense_threshold": 0.1, + "l0_raw": 128.0 + }, + "seed1": { + "d_sae": 12288, + "d_in": 768, + "l0": 128, + "recon_r2": 0.9473797138717784, + "l1_per_token": 4.725544648044432, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dup_signed": { + "0.1": 12171, + "0.2": 358, + "0.3": 46, + "0.4": 13, + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.1": 1, + "0.2": 113, + "0.3": 11, + "0.4": 4, + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.1": 12171, + "0.2": 60, + "0.3": 15, + "0.4": 6, + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.15287448465824127, + "n_dup_abs": { + "0.1": 12186, + "0.2": 605, + "0.3": 77, + "0.4": 22, + "0.5": 4, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.1": 1, + "0.2": 188, + "0.3": 15, + "0.4": 5, + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.1": 12186, + "0.2": 127, + "0.3": 39, + "0.4": 13, + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.16067351400852203, + "n_dense": 240, + "dense_rate": 0.01953125, + "dense_threshold": 0.1, + "l0_raw": 128.0 + }, + "seed2": { + "d_sae": 12288, + "d_in": 768, + "l0": 128, + "recon_r2": 0.9473060144759621, + "l1_per_token": 4.7358192101744505, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dup_signed": { + "0.1": 12166, + "0.2": 352, + "0.3": 50, + "0.4": 10, + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.1": 1, + "0.2": 114, + "0.3": 6, + "0.4": 3, + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.1": 12166, + "0.2": 89, + "0.3": 25, + "0.4": 6, + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.15283523499965668, + "n_dup_abs": { + "0.1": 12181, + "0.2": 583, + "0.3": 66, + "0.4": 13, + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.1": 1, + "0.2": 170, + "0.3": 12, + "0.4": 2, + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.1": 12181, + "0.2": 111, + "0.3": 44, + "0.4": 11, + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.16049137711524963, + "n_dense": 243, + "dense_rate": 0.019775390625, + "dense_threshold": 0.1, + "l0_raw": 128.0 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.46430912613868713, + "dcs_abs": 0.7022993564605713, + "frac_sign_flipped": 0.4956868489583333, + "alive_source": "stats", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.1": 0.4587426781654358, + "0.2": 0.46365466713905334, + "0.3": 0.4641101062297821, + "0.4": 0.4642353057861328, + "0.5": 0.4643448293209076, + "0.6": 0.46430912613868713, + "0.7": 0.46430912613868713, + "0.8": 0.46430912613868713, + "0.9": 0.46430912613868713 + }, + "unique_frac_signed": { + "0.1": 0.009562174479166666, + "0.2": 0.9799397786458334, + "0.3": 0.9969482421875, + "0.4": 0.999267578125, + "0.5": 0.9999593098958334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.1": 0.004386577529658098, + "0.2": 0.4543536518843515, + "0.3": 0.46269375458723516, + "0.4": 0.46389528969302773, + "0.5": 0.46432593508143327, + "0.6": 0.46430912613868713, + "0.7": 0.46430912613868713, + "0.8": 0.46430912613868713, + "0.9": 0.46430912613868713 + }, + "matched_frac_signed": { + "0.1": 0.006673177083333333, + "0.2": 0.6575520833333334, + "0.3": 0.5655110677083334, + "0.4": 0.5008138020833334, + "0.5": 0.4410807291666667, + "0.6": 0.3782552083333333, + "0.7": 0.304931640625, + "0.8": 0.21516927083333334, + "0.9": 0.09016927083333333 + }, + "matched_cos_abs": { + "0.1": 0.7257933616638184, + "0.2": 0.6972552537918091, + "0.3": 0.7019635438919067, + "0.4": 0.7023240923881531, + "0.5": 0.702376663684845, + "0.6": 0.7022993564605713, + "0.7": 0.7022993564605713, + "0.8": 0.7022993564605713, + "0.9": 0.7022993564605713 + }, + "unique_frac_abs": { + "0.1": 0.0084228515625, + "0.2": 0.96630859375, + "0.3": 0.9951578776041666, + "0.4": 0.9987386067708334, + "0.5": 0.9999186197916666, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.1": 0.00611324975034222, + "0.2": 0.6737637437763624, + "0.3": 0.6985645504949691, + "0.4": 0.701438185533334, + "0.5": 0.7023195041256258, + "0.6": 0.7022993564605713, + "0.7": 0.7022993564605713, + "0.8": 0.7022993564605713, + "0.9": 0.7022993564605713 + }, + "matched_frac_abs": { + "0.1": 0.0078125, + "0.2": 0.9140625, + "0.3": 0.911376953125, + "0.4": 0.87451171875, + "0.5": 0.8212076822916666, + "0.6": 0.73681640625, + "0.7": 0.608154296875, + "0.8": 0.4359537760416667, + "0.9": 0.17569986979166666 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.4672669470310211, + "dcs_abs": 0.6994244456291199, + "frac_sign_flipped": 0.4911295572916667, + "alive_source": "stats", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.1": 0.4353242814540863, + "0.2": 0.46695467829704285, + "0.3": 0.4671008288860321, + "0.4": 0.4671774208545685, + "0.5": 0.4672669470310211, + "0.6": 0.4672669470310211, + "0.7": 0.4672669470310211, + "0.8": 0.4672669470310211, + "0.9": 0.4672669470310211 + }, + "unique_frac_signed": { + "0.1": 0.009765625, + "0.2": 0.980224609375, + "0.3": 0.99658203125, + "0.4": 0.9993489583333334, + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.1": 0.0042512136860750616, + "0.2": 0.4577204671295476, + "0.3": 0.46550429284980055, + "0.4": 0.4668732688878663, + "0.5": 0.4672669470310211, + "0.6": 0.4672669470310211, + "0.7": 0.4672669470310211, + "0.8": 0.4672669470310211, + "0.9": 0.4672669470310211 + }, + "matched_frac_signed": { + "0.1": 0.006266276041666667, + "0.2": 0.6638997395833334, + "0.3": 0.5743001302083334, + "0.4": 0.5049641927083334, + "0.5": 0.4442545572916667, + "0.6": 0.3831380208333333, + "0.7": 0.3042805989583333, + "0.8": 0.21842447916666666, + "0.9": 0.08748372395833333 + }, + "matched_cos_abs": { + "0.1": 0.7672093510627747, + "0.2": 0.6937437653541565, + "0.3": 0.6991376876831055, + "0.4": 0.6995258927345276, + "0.5": 0.6994244456291199, + "0.6": 0.6994244456291199, + "0.7": 0.6994244456291199, + "0.8": 0.6994244456291199, + "0.9": 0.6994244456291199 + }, + "unique_frac_abs": { + "0.1": 0.008626302083333334, + "0.2": 0.9664713541666666, + "0.3": 0.9954833984375, + "0.4": 0.9989827473958334, + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.1": 0.006618179623425628, + "0.2": 0.6704834763465138, + "0.3": 0.6959799613105133, + "0.4": 0.6988142981984614, + "0.5": 0.6994244456291199, + "0.6": 0.6994244456291199, + "0.7": 0.6994244456291199, + "0.8": 0.6994244456291199, + "0.9": 0.6994244456291199 + }, + "matched_frac_abs": { + "0.1": 0.008138020833333334, + "0.2": 0.9117838541666666, + "0.3": 0.9112955729166666, + "0.4": 0.8724772135416666, + "0.5": 0.8211263020833334, + "0.6": 0.7334798177083334, + "0.7": 0.5975748697916666, + "0.8": 0.4273274739583333, + "0.9": 0.172607421875 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.4624062478542328, + "dcs_abs": 0.7074873447418213, + "frac_sign_flipped": 0.5096028645833334, + "alive_source": "stats", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.1": 0.47585415840148926, + "0.2": 0.46254146099090576, + "0.3": 0.46251624822616577, + "0.4": 0.4624234735965729, + "0.5": 0.46243977546691895, + "0.6": 0.4624062478542328, + "0.7": 0.4624062478542328, + "0.8": 0.4624062478542328, + "0.9": 0.4624062478542328 + }, + "unique_frac_signed": { + "0.1": 0.009806315104166666, + "0.2": 0.9803466796875, + "0.3": 0.9967854817708334, + "0.4": 0.9993489583333334, + "0.5": 0.9999593098958334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.1": 0.004666375820913041, + "0.2": 0.45345098550023977, + "0.3": 0.461029481314957, + "0.4": 0.4621224166476168, + "0.5": 0.4624209587442844, + "0.6": 0.4624062478542328, + "0.7": 0.4624062478542328, + "0.8": 0.4624062478542328, + "0.9": 0.4624062478542328 + }, + "matched_frac_signed": { + "0.1": 0.006673177083333333, + "0.2": 0.6542154947916666, + "0.3": 0.5621744791666666, + "0.4": 0.4906412760416667, + "0.5": 0.4315592447916667, + "0.6": 0.3750813802083333, + "0.7": 0.3033854166666667, + "0.8": 0.22151692708333334, + "0.9": 0.093017578125 + }, + "matched_cos_abs": { + "0.1": 0.7786716222763062, + "0.2": 0.7031784653663635, + "0.3": 0.7071307897567749, + "0.4": 0.7075624465942383, + "0.5": 0.7075602412223816, + "0.6": 0.7074873447418213, + "0.7": 0.7074873447418213, + "0.8": 0.7074873447418213, + "0.9": 0.7074873447418213 + }, + "unique_frac_abs": { + "0.1": 0.008585611979166666, + "0.2": 0.9662272135416666, + "0.3": 0.9952799479166666, + "0.4": 0.9988606770833334, + "0.5": 0.9999186197916666, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.1": 0.006685372408052595, + "0.2": 0.6794301692134468, + "0.3": 0.7037930955993943, + "0.4": 0.7067563044838607, + "0.5": 0.7075026598225425, + "0.6": 0.7074873447418213, + "0.7": 0.7074873447418213, + "0.8": 0.7074873447418213, + "0.9": 0.7074873447418213 + }, + "matched_frac_abs": { + "0.1": 0.008138020833333334, + "0.2": 0.912841796875, + "0.3": 0.912109375, + "0.4": 0.8779296875, + "0.5": 0.826904296875, + "0.6": 0.7451985677083334, + "0.7": 0.6197916666666666, + "0.8": 0.4498697916666667, + "0.9": 0.18782552083333334 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 241, + "n_dense_std": 1.7320508075688772, + "l0_mean": 128, + "l0_std": 0.0, + "recon_r2_mean": 0.9473207226558779, + "recon_r2_std": null, + "l1_per_token_mean": 4.733107309611182, + "l1_per_token_std": 0.0066361938244437835, + "babel_mu1_signed_mean": 0.15281703571478525, + "babel_mu1_signed_std": 6.838950768620448e-05, + "n_dup_signed_per_threshold": { + "0.1": 12169.666666666666, + "0.2": 357.0, + "0.3": 49.0, + "0.4": 11.333333333333334, + "0.5": 0.6666666666666666, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_signed_band": 0.0, + "n_dup_signed_band_std": 0.0, + "n_dup_clusters_signed_per_threshold": { + "0.1": 1.0, + "0.2": 113.33333333333333, + "0.3": 9.333333333333334, + "0.4": 3.0, + "0.5": 0.3333333333333333, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_signed_band": 0.0, + "n_dup_clusters_signed_band_std": 0.0, + "largest_dup_signed_per_threshold": { + "0.1": 12169.666666666666, + "0.2": 67.66666666666667, + "0.3": 20.333333333333332, + "0.4": 6.333333333333333, + "0.5": 1.3333333333333333, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_signed_band": 1.0, + "largest_dup_signed_band_std": 0.0, + "dcs_signed_mean": 0.46466077367464703, + "dcs_signed_std": 0.002449355244542465, + "dedup_dcs_signed_per_threshold": { + "0.1": 0.004434722345548733, + "0.2": 0.45517503483804633, + "0.3": 0.46307584291733095, + "0.4": 0.46429699174283695, + "0.5": 0.46467128028557964, + "0.6": 0.46466077367464703, + "0.7": 0.46466077367464703, + "0.8": 0.46466077367464703, + "0.9": 0.46466077367464703 + }, + "dedup_dcs_signed_band": 0.46466077367464703, + "dedup_dcs_signed_band_std": 0.002449355244542465, + "matched_cos_signed_per_threshold": { + "0.1": 0.45664037267367047, + "0.2": 0.464383602142334, + "0.3": 0.46457572778066, + "0.4": 0.46461206674575806, + "0.5": 0.46468385060628253, + "0.6": 0.46466077367464703, + "0.7": 0.46466077367464703, + "0.8": 0.46466077367464703, + "0.9": 0.46466077367464703 + }, + "matched_cos_signed_band": 0.46466077367464703, + "matched_cos_signed_band_std": 0.002449355244542465, + "unique_frac_signed_per_threshold": { + "0.1": 0.009711371527777776, + "0.2": 0.9801703559027778, + "0.3": 0.9967719184027778, + "0.4": 0.9993218315972223, + "0.5": 0.999972873263889, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_signed_band": 1.0, + "unique_frac_signed_band_std": 0.0, + "matched_frac_signed_per_threshold": { + "0.1": 0.006537543402777777, + "0.2": 0.6585557725694445, + "0.3": 0.5673285590277778, + "0.4": 0.49880642361111116, + "0.5": 0.43896484375, + "0.6": 0.3788248697916667, + "0.7": 0.30419921875, + "0.8": 0.21837022569444445, + "0.9": 0.09022352430555554 + }, + "matched_frac_signed_band": 0.24790445963541669, + "matched_frac_signed_band_std": 0.0006707699186944385, + "babel_mu1_abs_mean": 0.16046898563702902, + "babel_mu1_abs_std": 0.00021659391870105051, + "n_dup_abs_per_threshold": { + "0.1": 12184.0, + "0.2": 594.0, + "0.3": 71.66666666666667, + "0.4": 17.0, + "0.5": 1.3333333333333333, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_abs_band": 0.0, + "n_dup_abs_band_std": 0.0, + "n_dup_clusters_abs_per_threshold": { + "0.1": 1.0, + "0.2": 180.33333333333334, + "0.3": 14.0, + "0.4": 3.0, + "0.5": 0.6666666666666666, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_abs_band": 0.0, + "n_dup_clusters_abs_band_std": 0.0, + "largest_dup_abs_per_threshold": { + "0.1": 12184.0, + "0.2": 121.33333333333333, + "0.3": 40.666666666666664, + "0.4": 12.666666666666666, + "0.5": 1.3333333333333333, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_abs_band": 1.0, + "largest_dup_abs_band_std": 0.0, + "dcs_abs_mean": 0.7030703822771708, + "dcs_abs_std": 0.00408637322485325, + "dedup_dcs_abs_per_threshold": { + "0.1": 0.006472267260606815, + "0.2": 0.6745591297787743, + "0.3": 0.6994458691349589, + "0.4": 0.7023362627385521, + "0.5": 0.7030822031924293, + "0.6": 0.7030703822771708, + "0.7": 0.7030703822771708, + "0.8": 0.7030703822771708, + "0.9": 0.7030703822771708 + }, + "dedup_dcs_abs_band": 0.7030703822771708, + "dedup_dcs_abs_band_std": 0.00408637322485325, + "matched_cos_abs_per_threshold": { + "0.1": 0.7572247783342997, + "0.2": 0.6980591615041097, + "0.3": 0.7027440071105957, + "0.4": 0.703137477238973, + "0.5": 0.7031204501787821, + "0.6": 0.7030703822771708, + "0.7": 0.7030703822771708, + "0.8": 0.7030703822771708, + "0.9": 0.7030703822771708 + }, + "matched_cos_abs_band": 0.7030703822771708, + "matched_cos_abs_band_std": 0.00408637322485325, + "unique_frac_abs_per_threshold": { + "0.1": 0.008544921875, + "0.2": 0.966335720486111, + "0.3": 0.9953070746527777, + "0.4": 0.9988606770833334, + "0.5": 0.9999457465277777, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_abs_band": 1.0, + "unique_frac_abs_band_std": 0.0, + "matched_frac_abs_per_threshold": { + "0.1": 0.00802951388888889, + "0.2": 0.9128960503472222, + "0.3": 0.9115939670138888, + "0.4": 0.8749728732638888, + "0.5": 0.8230794270833334, + "0.6": 0.738498263888889, + "0.7": 0.6085069444444443, + "0.8": 0.4377170138888889, + "0.9": 0.1787109375 + }, + "matched_frac_abs_band": 0.4908582899305555, + "matched_frac_abs_band_std": 0.009082427143445183, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/config.json b/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d1a4836058c29cfdbfdadd94e9d4c640e079a7be --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 128, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/lock.lock b/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..e9bf12ccfb11e2697b5af73f8a731c354434630b --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 127.7522, + "l0_per_token_0.05": 19.707093603515624, + "l0_per_token_0.1": 5.5419546875, + "l0_per_token_0.5": 0.37530859375, + "l0_per_token_1.0": 0.001928271484375, + "l1_per_token": 4.737958070614665, + "r2_score": 0.9472764396198932, + "l2_loss": 0.03510181688132131 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/config.json b/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..28bd0a9a09fabf165083e2ee78027182c7ba5be3 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 128, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/lock.lock b/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..4d76300d18adfc58f3c66373fd3e49708854da16 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 127.73053618164063, + "l0_per_token_0.05": 19.58694951171875, + "l0_per_token_0.1": 5.489561279296875, + "l0_per_token_0.5": 0.376380517578125, + "l0_per_token_1.0": 0.00172236328125, + "l1_per_token": 4.725544648044432, + "r2_score": 0.9473797138717784, + "l2_loss": 0.0350330598806162 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/config.json b/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..dc9d411495e2547d51bacace5eb57c467f1a668e --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 128, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/lock.lock b/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..bd41d604622cf87946cd3a1e7ab2afda21dc68a9 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 127.72996455078125, + "l0_per_token_0.05": 19.639303173828125, + "l0_per_token_0.1": 5.5228298828125, + "l0_per_token_0.5": 0.377465478515625, + "l0_per_token_1.0": 0.001918603515625, + "l1_per_token": 4.7358192101744505, + "r2_score": 0.9473060144759621, + "l2_loss": 0.03508212680017851 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/code_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x16/l0_32/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..161dfcafc48d86ca192b27aac2f15011f3633814 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 32.0, + "recon_r2": 0.9016975595767329, + "l1_per_token": 2.5368724051361653 + }, + "seed1": { + "l0": 32.0, + "recon_r2": 0.9018428926849782, + "l1_per_token": 2.5332101443712522 + }, + "seed2": { + "l0": 32.0, + "recon_r2": 0.901553944275022, + "l1_per_token": 2.537756573778936 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.308376699254761, + "mcc_abs": 0.3126719792052181, + "act_iou": 0.21962694373929656, + "act_iou_n_pairs": 10648, + "input_r2": 0.968782198458154, + "matched_mcc_signed": { + "0.6": 0.8064855326621341, + "0.7": 0.843817904218046, + "0.8": 0.8758067424555736, + "0.9": 0.9074249118690291 + }, + "matched_mcc_abs": { + "0.6": 0.8338038047254341, + "0.7": 0.8547343081386967, + "0.8": 0.8793657237555296, + "0.9": 0.9078189152316456 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.30234420689913005, + "mcc_abs": 0.3062274086667045, + "act_iou": 0.21785790303447647, + "act_iou_n_pairs": 10447, + "input_r2": 0.9682132708902778, + "matched_mcc_signed": { + "0.6": 0.8055498943549442, + "0.7": 0.8385526136108705, + "0.8": 0.8752817398329477, + "0.9": 0.9057537116744127 + }, + "matched_mcc_abs": { + "0.6": 0.8349890540681039, + "0.7": 0.8534548795999692, + "0.8": 0.8788221036407499, + "0.9": 0.9071050158362797 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.30512430748097047, + "mcc_abs": 0.3093440208141284, + "act_iou": 0.21478136497402517, + "act_iou_n_pairs": 10656, + "input_r2": 0.9675907712748805, + "matched_mcc_signed": { + "0.6": 0.8015980598423272, + "0.7": 0.8376520189532873, + "0.8": 0.8719231862298532, + "0.9": 0.906375672330228 + }, + "matched_mcc_abs": { + "0.6": 0.8317022236624306, + "0.7": 0.8529146701584874, + "0.8": 0.8776759701211921, + "0.9": 0.906653296250762 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 32.0, + "l0_std": 0.0, + "recon_r2_mean": 0.901698132178911, + "recon_r2_std": 0.00014447505600965715, + "l1_per_token_mean": 2.5359463744287845, + "l1_per_token_std": 0.002410529971400832, + "input_r2_mean": 0.968195413541104, + "input_r2_std": 0.0005959142949707115, + "mcc_signed_mean": 0.3052817378782872, + "mcc_signed_std": 0.0030193259599927253, + "mcc_abs_mean": 0.309414469562017, + "mcc_abs_std": 0.0032228627997525263, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8045444956198019, + "0.7": 0.8400075122607347, + "0.8": 0.8743372228394581, + "0.9": 0.9065180986245566 + }, + "matched_mcc_signed_band": 0.8563518323361379, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8334983608186562, + "0.7": 0.8537012859657178, + "0.8": 0.8786212658391572, + "0.9": 0.9071924091062291 + }, + "matched_mcc_abs_band": 0.8682533304324401, + "act_iou_mean": 0.2174220705825994, + "act_iou_std": 0.0024520136290800136, + "act_iou_n_pairs": 10583.666666666666, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/dict_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x16/l0_32/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..c6965c51f61981e3ad11b058ff2414a2f915a32b --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 12288, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.902168522404218, + "l1_per_token": 2.5336257038867065, + "stats_source": "stats", + "n_dead": 2739, + "dead_rate": 0.222900390625, + "alive_source": "stats", + "n_dense": 16, + "dense_rate": 0.0013020833333333333, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1157, + "0.6": 714, + "0.7": 386, + "0.8": 127, + "0.9": 23 + }, + "n_dup_clusters_signed": { + "0.5": 516, + "0.6": 331, + "0.7": 183, + "0.8": 61, + "0.9": 11 + }, + "largest_dup_signed": { + "0.5": 9, + "0.6": 6, + "0.7": 5, + "0.8": 3, + "0.9": 3 + }, + "babel_mu1_signed": 0.2650097906589508, + "n_dup_abs": { + "0.5": 1472, + "0.6": 940, + "0.7": 530, + "0.8": 212, + "0.9": 42 + }, + "n_dup_clusters_abs": { + "0.5": 600, + "0.6": 408, + "0.7": 241, + "0.8": 99, + "0.9": 20 + }, + "largest_dup_abs": { + "0.5": 14, + "0.6": 9, + "0.7": 5, + "0.8": 4, + "0.9": 4 + }, + "babel_mu1_abs": 0.29133734107017517 + }, + "seed1": { + "d_sae": 12288, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.9023354724249185, + "l1_per_token": 2.529936031870183, + "stats_source": "stats", + "n_dead": 2350, + "dead_rate": 0.19124348958333334, + "alive_source": "stats", + "n_dense": 14, + "dense_rate": 0.0011393229166666667, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1149, + "0.6": 736, + "0.7": 400, + "0.8": 152, + "0.9": 21 + }, + "n_dup_clusters_signed": { + "0.5": 497, + "0.6": 333, + "0.7": 188, + "0.8": 73, + "0.9": 10 + }, + "largest_dup_signed": { + "0.5": 9, + "0.6": 5, + "0.7": 4, + "0.8": 4, + "0.9": 3 + }, + "babel_mu1_signed": 0.26779136061668396, + "n_dup_abs": { + "0.5": 1554, + "0.6": 997, + "0.7": 551, + "0.8": 228, + "0.9": 40 + }, + "n_dup_clusters_abs": { + "0.5": 631, + "0.6": 431, + "0.7": 250, + "0.8": 106, + "0.9": 19 + }, + "largest_dup_abs": { + "0.5": 16, + "0.6": 6, + "0.7": 5, + "0.8": 4, + "0.9": 3 + }, + "babel_mu1_abs": 0.2961384952068329 + }, + "seed2": { + "d_sae": 12288, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.9020171694312822, + "l1_per_token": 2.5345750532533895, + "stats_source": "stats", + "n_dead": 2765, + "dead_rate": 0.22501627604166666, + "alive_source": "stats", + "n_dense": 16, + "dense_rate": 0.0013020833333333333, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1324, + "0.6": 785, + "0.7": 423, + "0.8": 167, + "0.9": 42 + }, + "n_dup_clusters_signed": { + "0.5": 582, + "0.6": 356, + "0.7": 198, + "0.8": 79, + "0.9": 18 + }, + "largest_dup_signed": { + "0.5": 8, + "0.6": 5, + "0.7": 4, + "0.8": 4, + "0.9": 4 + }, + "babel_mu1_signed": 0.27672192454338074, + "n_dup_abs": { + "0.5": 1735, + "0.6": 1100, + "0.7": 635, + "0.8": 258, + "0.9": 61 + }, + "n_dup_clusters_abs": { + "0.5": 705, + "0.6": 475, + "0.7": 288, + "0.8": 121, + "0.9": 28 + }, + "largest_dup_abs": { + "0.5": 9, + "0.6": 8, + "0.7": 6, + "0.8": 5, + "0.9": 3 + }, + "babel_mu1_abs": 0.30721229314804077 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.48426708579063416, + "dcs_abs": 0.704206645488739, + "frac_sign_flipped": 0.5008951822916666, + "alive_source": "stats", + "n_alive_a": 9549, + "n_alive_b": 9938, + "matched_cos_signed": { + "0.5": 0.515743613243103, + "0.6": 0.5147678852081299, + "0.7": 0.5140931010246277, + "0.8": 0.513370156288147, + "0.9": 0.5130211114883423 + }, + "unique_frac_signed": { + "0.5": 0.9797300764612306, + "0.6": 0.9868630368963924, + "0.7": 0.9926617745163442, + "0.8": 0.9975368194180736, + "0.9": 0.9997947349515062 + }, + "dedup_dcs_signed": { + "0.5": 0.5052895296370566, + "0.6": 0.5080053984932286, + "0.7": 0.5103205699297171, + "0.8": 0.5121056328878375, + "0.9": 0.5129158061850143 + }, + "matched_frac_signed": { + "0.5": 0.4465541129984092, + "0.6": 0.4125827474726741, + "0.7": 0.3714271052496536, + "0.8": 0.3234977164263355, + "0.9": 0.25114178683224714 + }, + "matched_cos_abs": { + "0.5": 0.7940940260887146, + "0.6": 0.7909408807754517, + "0.7": 0.7890170812606812, + "0.8": 0.7869129180908203, + "0.9": 0.7863107919692993 + }, + "unique_frac_abs": { + "0.5": 0.9692102427259198, + "0.6": 0.9801919228203417, + "0.7": 0.9888643711192077, + "0.8": 0.9962025966028635, + "0.9": 0.9997434186893827 + }, + "dedup_dcs_abs": { + "0.5": 0.7696440637726459, + "0.6": 0.7752738627645046, + "0.7": 0.7802308798631562, + "0.8": 0.7839246923024116, + "0.9": 0.7861090393157433 + }, + "matched_frac_abs": { + "0.5": 0.8162364653358649, + "0.6": 0.785138810489044, + "0.7": 0.7324883255503669, + "0.8": 0.6443269872222507, + "0.9": 0.502694103761482 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.48269525170326233, + "dcs_abs": 0.7041468024253845, + "frac_sign_flipped": 0.5016276041666666, + "alive_source": "stats", + "n_alive_a": 9549, + "n_alive_b": 9523, + "matched_cos_signed": { + "0.5": 0.5082270503044128, + "0.6": 0.5068737864494324, + "0.7": 0.5056464672088623, + "0.8": 0.5045716166496277, + "0.9": 0.5042540431022644 + }, + "unique_frac_signed": { + "0.5": 0.9826971476510067, + "0.6": 0.9886744966442953, + "0.7": 0.9934458892617449, + "0.8": 0.9978502516778524, + "0.9": 0.9996329697986577 + }, + "dedup_dcs_signed": { + "0.5": 0.4994332726932312, + "0.6": 0.5011331856800806, + "0.7": 0.502332404268368, + "0.8": 0.5034869146633318, + "0.9": 0.5040689666392969 + }, + "matched_frac_signed": { + "0.5": 0.447881711409396, + "0.6": 0.411493288590604, + "0.7": 0.37258808724832215, + "0.8": 0.32298657718120805, + "0.9": 0.24611996644295303 + }, + "matched_cos_abs": { + "0.5": 0.7914395332336426, + "0.6": 0.7874656915664673, + "0.7": 0.7836632132530212, + "0.8": 0.781852126121521, + "0.9": 0.7807745933532715 + }, + "unique_frac_abs": { + "0.5": 0.9721581375838926, + "0.6": 0.9812814597315436, + "0.7": 0.9891463926174496, + "0.8": 0.9962248322147651, + "0.9": 0.9995281040268457 + }, + "dedup_dcs_abs": { + "0.5": 0.7694043826386833, + "0.6": 0.7727254833088525, + "0.7": 0.7751576404162251, + "0.8": 0.7789005031621696, + "0.9": 0.7804061489667269 + }, + "matched_frac_abs": { + "0.5": 0.8282298657718121, + "0.6": 0.79435822147651, + "0.7": 0.744861577181208, + "0.8": 0.6577181208053692, + "0.9": 0.50251677852349 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.486629456281662, + "dcs_abs": 0.7040589451789856, + "frac_sign_flipped": 0.4988606770833333, + "alive_source": "stats", + "n_alive_a": 9938, + "n_alive_b": 9523, + "matched_cos_signed": { + "0.5": 0.5160989761352539, + "0.6": 0.5149685144424438, + "0.7": 0.514082670211792, + "0.8": 0.513390302658081, + "0.9": 0.5131831169128418 + }, + "unique_frac_signed": { + "0.5": 0.9794974564513642, + "0.6": 0.9865371769179384, + "0.7": 0.9927033554288063, + "0.8": 0.9975849134165767, + "0.9": 0.9996403062535327 + }, + "dedup_dcs_signed": { + "0.5": 0.5055176344016346, + "0.6": 0.5080355844396731, + "0.7": 0.5103315916870463, + "0.8": 0.512150420626072, + "0.9": 0.5129985281548957 + }, + "matched_frac_signed": { + "0.5": 0.44550639741020504, + "0.6": 0.4090231745542367, + "0.7": 0.36781254817326964, + "0.8": 0.32053851292328245, + "0.9": 0.24130311905863008 + }, + "matched_cos_abs": { + "0.5": 0.7925868034362793, + "0.6": 0.7894279360771179, + "0.7": 0.786578357219696, + "0.8": 0.7848519682884216, + "0.9": 0.7842956185340881 + }, + "unique_frac_abs": { + "0.5": 0.9677817172807153, + "0.6": 0.9791891475258209, + "0.7": 0.9892091876059812, + "0.8": 0.9958892143260881, + "0.9": 0.9995889214326088 + }, + "dedup_dcs_abs": { + "0.5": 0.7670510177235951, + "0.6": 0.7729992677604213, + "0.7": 0.7780905377337428, + "0.8": 0.78162561006104, + "0.9": 0.7839732114148099 + }, + "matched_frac_abs": { + "0.5": 0.8134217152253225, + "0.6": 0.7778634191459842, + "0.7": 0.722470582190021, + "0.8": 0.6400493294280869, + "0.9": 0.49062227018138843 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 2618, + "n_dead_std": 232.4585984643287, + "n_dense_mean": 15, + "n_dense_std": 1.1547005383792517, + "l0_mean": 32, + "l0_std": 0.0, + "recon_r2_mean": 0.9021737214201395, + "recon_r2_std": null, + "l1_per_token_mean": 2.5327122630034267, + "l1_per_token_std": 0.0024506959284858612, + "babel_mu1_signed_mean": 0.2698410252730052, + "babel_mu1_signed_std": 0.006119180005052326, + "n_dup_signed_per_threshold": { + "0.5": 1210.0, + "0.6": 745.0, + "0.7": 403.0, + "0.8": 148.66666666666666, + "0.9": 28.666666666666668 + }, + "n_dup_signed_band": 331.33333333333337, + "n_dup_signed_band_std": 21.17240735800569, + "n_dup_clusters_signed_per_threshold": { + "0.5": 531.6666666666666, + "0.6": 340.0, + "0.7": 189.66666666666666, + "0.8": 71.0, + "0.9": 13.0 + }, + "n_dup_clusters_signed_band": 153.41666666666666, + "n_dup_clusters_signed_band_std": 8.39022248413791, + "largest_dup_signed_per_threshold": { + "0.5": 8.666666666666666, + "0.6": 5.333333333333333, + "0.7": 4.333333333333333, + "0.8": 3.6666666666666665, + "0.9": 3.3333333333333335 + }, + "largest_dup_signed_band": 4.166666666666666, + "largest_dup_signed_band_std": 0.14433756729740646, + "dcs_signed_mean": 0.48453059792518616, + "dcs_signed_std": 0.001980295533491737, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.5034134789106408, + "0.6": 0.5057247228709941, + "0.7": 0.5076615219617104, + "0.8": 0.5092476560590805, + "0.9": 0.5099944336597356 + }, + "dedup_dcs_signed_band": 0.5081570836378801, + "dedup_dcs_signed_band_std": 0.004678070667047931, + "matched_cos_signed_per_threshold": { + "0.5": 0.5133565465609232, + "0.6": 0.5122033953666687, + "0.7": 0.5112740794817606, + "0.8": 0.5104440251986185, + "0.9": 0.5101527571678162 + }, + "matched_cos_signed_band": 0.511018564303716, + "matched_cos_signed_band_std": 0.004921050892902177, + "unique_frac_signed_per_threshold": { + "0.5": 0.9806415601878671, + "0.6": 0.987358236819542, + "0.7": 0.9929370064022983, + "0.8": 0.9976573281708343, + "0.9": 0.9996893370012322 + }, + "unique_frac_signed_band": 0.9944104770984767, + "unique_frac_signed_band_std": 0.0004275176874978255, + "matched_frac_signed_per_threshold": { + "0.5": 0.4466474072726701, + "0.6": 0.4110330702058383, + "0.7": 0.37060924689041513, + "0.8": 0.3223409355102753, + "0.9": 0.24618829077794344 + }, + "matched_frac_signed_band": 0.337542885846118, + "matched_frac_signed_band_std": 0.0025805050642128033, + "babel_mu1_abs_mean": 0.2982293764750163, + "babel_mu1_abs_std": 0.008141398173954059, + "n_dup_abs_per_threshold": { + "0.5": 1587.0, + "0.6": 1012.3333333333334, + "0.7": 572.0, + "0.8": 232.66666666666666, + "0.9": 47.666666666666664 + }, + "n_dup_abs_band": 466.16666666666674, + "n_dup_abs_band_std": 42.57444460393269, + "n_dup_clusters_abs_per_threshold": { + "0.5": 645.3333333333334, + "0.6": 438.0, + "0.7": 259.6666666666667, + "0.8": 108.66666666666667, + "0.9": 22.333333333333332 + }, + "n_dup_clusters_abs_band": 207.16666666666669, + "n_dup_clusters_abs_band_std": 18.65699154026, + "largest_dup_abs_per_threshold": { + "0.5": 13.0, + "0.6": 7.666666666666667, + "0.7": 5.333333333333333, + "0.8": 4.333333333333333, + "0.9": 3.3333333333333335 + }, + "largest_dup_abs_band": 5.166666666666666, + "largest_dup_abs_band_std": 0.5773502691896258, + "dcs_abs_mean": 0.7041374643643697, + "dcs_abs_std": 7.4291620745217e-05, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.7686998213783082, + "0.6": 0.7736662046112595, + "0.7": 0.7778263526710414, + "0.8": 0.7814836018418737, + "0.9": 0.7834961332324267 + }, + "dedup_dcs_abs_band": 0.7791180730891504, + "dedup_dcs_abs_band_std": 0.002294065491483052, + "matched_cos_abs_per_threshold": { + "0.5": 0.7927067875862122, + "0.6": 0.7892781694730123, + "0.7": 0.7864195505777994, + "0.8": 0.784539004166921, + "0.9": 0.7837936679522196 + }, + "matched_cos_abs_band": 0.7860075980424881, + "matched_cos_abs_band_std": 0.002440408549049815, + "unique_frac_abs_per_threshold": { + "0.5": 0.9697166991968426, + "0.6": 0.9802208433592354, + "0.7": 0.9890733171142129, + "0.8": 0.9961055477145723, + "0.9": 0.9996201480496124 + }, + "unique_frac_abs_band": 0.9912549640594083, + "unique_frac_abs_band_std": 0.00028806476472473404, + "matched_frac_abs_per_threshold": { + "0.5": 0.8192960154443331, + "0.6": 0.7857868170371795, + "0.7": 0.7332734949738654, + "0.8": 0.6473648124852356, + "0.9": 0.4986110508221202 + }, + "matched_frac_abs_band": 0.6662590438296001, + "matched_frac_abs_band_std": 0.008556549389740924, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/config.json b/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..453661d3a2732a1758a572f2b2d5b55d9d69be6e --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 32, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/lock.lock b/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..9a2b99eaa0f8c44d99c5bbef3515f43ee2d79dd4 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 31.988513330078124, + "l0_per_token_0.05": 17.016170654296875, + "l0_per_token_0.1": 5.48449921875, + "l0_per_token_0.5": 0.362630029296875, + "l0_per_token_1.0": 0.017250927734375, + "l1_per_token": 2.5336257038867065, + "r2_score": 0.902168522404218, + "l2_loss": 0.06513335948935535 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/config.json b/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ac5d516faa65f1748a0ccabb4d1d287f5a2b30c8 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 32, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/lock.lock b/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..03e76fa94898fcfb6e58461e0cf10e3dfc1929c8 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 31.983575634765625, + "l0_per_token_0.05": 16.96002568359375, + "l0_per_token_0.1": 5.46407265625, + "l0_per_token_0.5": 0.363770263671875, + "l0_per_token_1.0": 0.01745458984375, + "l1_per_token": 2.529936031870183, + "r2_score": 0.9023354724249185, + "l2_loss": 0.065022209009139 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/config.json b/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..06369c818260ca5170c0dfe7473da75248c01f80 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 32, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/lock.lock b/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..3bdd9119b26cbe5a82181f5b984dd559d1c793f7 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 31.98508466796875, + "l0_per_token_0.05": 17.0400123046875, + "l0_per_token_0.1": 5.496093212890625, + "l0_per_token_0.5": 0.36267294921875, + "l0_per_token_1.0": 0.017344384765625, + "l1_per_token": 2.5345750532533895, + "r2_score": 0.9020171694312822, + "l2_loss": 0.0652341259076725 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/code_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x16/l0_64/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..1fa55926d0222795d6fc512147290e72feb3976d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 64.0, + "recon_r2": 0.925350156888914, + "l1_per_token": 3.4015277292469586 + }, + "seed1": { + "l0": 64.0, + "recon_r2": 0.9254350853685667, + "l1_per_token": 3.391007159912762 + }, + "seed2": { + "l0": 64.0, + "recon_r2": 0.9253811098136913, + "l1_per_token": 3.399506216056933 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.4207209365749514, + "mcc_abs": 0.4271845279673408, + "act_iou": 0.23072858947173316, + "act_iou_n_pairs": 12288, + "input_r2": 0.9674134905008003, + "matched_mcc_signed": { + "0.6": 0.7989614232084052, + "0.7": 0.8248976160274838, + "0.8": 0.8515033983012729, + "0.9": 0.8876085435436981 + }, + "matched_mcc_abs": { + "0.6": 0.8027244954331271, + "0.7": 0.8244098459782976, + "0.8": 0.8511560015685371, + "0.9": 0.8877101869649973 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.4195708145442103, + "mcc_abs": 0.4268452563859899, + "act_iou": 0.23171477261538032, + "act_iou_n_pairs": 12288, + "input_r2": 0.9666262567022783, + "matched_mcc_signed": { + "0.6": 0.7973204039501537, + "0.7": 0.8237260682901125, + "0.8": 0.8523825586122493, + "0.9": 0.8894758225385236 + }, + "matched_mcc_abs": { + "0.6": 0.8033626434836594, + "0.7": 0.8243614255384568, + "0.8": 0.8531013218271484, + "0.9": 0.889370157227176 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.41907692488615683, + "mcc_abs": 0.42557184878713733, + "act_iou": 0.22966816616569483, + "act_iou_n_pairs": 12288, + "input_r2": 0.9680140194905632, + "matched_mcc_signed": { + "0.6": 0.7956352035651377, + "0.7": 0.8211987003915764, + "0.8": 0.8501320342809222, + "0.9": 0.8877780355453729 + }, + "matched_mcc_abs": { + "0.6": 0.8019756741391523, + "0.7": 0.824356748534688, + "0.8": 0.851921183476043, + "0.9": 0.8889276532222591 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 64.0, + "l0_std": 0.0, + "recon_r2_mean": 0.925388784023724, + "recon_r2_std": 4.2981179471356485e-05, + "l1_per_token_mean": 3.3973470350722175, + "l1_per_token_std": 0.005582753950558539, + "input_r2_mean": 0.9673512555645473, + "input_r2_std": 0.0006959714646484294, + "mcc_signed_mean": 0.4197895586684395, + "mcc_signed_std": 0.000843552222530949, + "mcc_abs_mean": 0.42653387771348933, + "mcc_abs_std": 0.0008502358748203901, + "matched_mcc_signed_per_threshold": { + "0.6": 0.7973056769078989, + "0.7": 0.8232741282363909, + "0.8": 0.8513393303981482, + "0.9": 0.8882874672091982 + }, + "matched_mcc_signed_band": 0.840051650687909, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8026876043519796, + "0.7": 0.824376006683814, + "0.8": 0.8520595022905763, + "0.9": 0.8886693324714775 + }, + "matched_mcc_abs_band": 0.8419481114494619, + "act_iou_mean": 0.2307038427509361, + "act_iou_std": 0.0010235276205927308, + "act_iou_n_pairs": 12288.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/dict_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x16/l0_64/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..fc44e87ec6e13e754e4477540a96f1061d741cde --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 12288, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.9257008624355846, + "l1_per_token": 3.3973256462022587, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 71, + "dense_rate": 0.005777994791666667, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.16171056032180786, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.17077326774597168 + }, + "seed1": { + "d_sae": 12288, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.925800207561149, + "l1_per_token": 3.3871114266376185, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 72, + "dense_rate": 0.005859375, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.16291916370391846, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.1724100112915039 + }, + "seed2": { + "d_sae": 12288, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.925717261910719, + "l1_per_token": 3.3956643372152553, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 76, + "dense_rate": 0.006184895833333333, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.16207580268383026, + "n_dup_abs": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.17149348556995392 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.49124541878700256, + "dcs_abs": 0.7684378027915955, + "frac_sign_flipped": 0.5047200520833334, + "alive_source": "stats", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.5": 0.49124541878700256, + "0.6": 0.49124541878700256, + "0.7": 0.49124541878700256, + "0.8": 0.49124541878700256, + "0.9": 0.49124541878700256 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.49124541878700256, + "0.6": 0.49124541878700256, + "0.7": 0.49124541878700256, + "0.8": 0.49124541878700256, + "0.9": 0.49124541878700256 + }, + "matched_frac_signed": { + "0.5": 0.455078125, + "0.6": 0.4130859375, + "0.7": 0.363525390625, + "0.8": 0.3006998697916667, + "0.9": 0.1923828125 + }, + "matched_cos_abs": { + "0.5": 0.7684378027915955, + "0.6": 0.7684378027915955, + "0.7": 0.7684378027915955, + "0.8": 0.7684378027915955, + "0.9": 0.7684378027915955 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7684378027915955, + "0.6": 0.7684378027915955, + "0.7": 0.7684378027915955, + "0.8": 0.7684378027915955, + "0.9": 0.7684378027915955 + }, + "matched_frac_abs": { + "0.5": 0.8655598958333334, + "0.6": 0.8142903645833334, + "0.7": 0.7362467447916666, + "0.8": 0.609619140625, + "0.9": 0.3881022135416667 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.49265971779823303, + "dcs_abs": 0.7701990604400635, + "frac_sign_flipped": 0.5032552083333334, + "alive_source": "stats", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.5": 0.49267175793647766, + "0.6": 0.49265971779823303, + "0.7": 0.49265971779823303, + "0.8": 0.49265971779823303, + "0.9": 0.49265971779823303 + }, + "unique_frac_signed": { + "0.5": 0.9999593098958334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.49265171107132727, + "0.6": 0.49265971779823303, + "0.7": 0.49265971779823303, + "0.8": 0.49265971779823303, + "0.9": 0.49265971779823303 + }, + "matched_frac_signed": { + "0.5": 0.459228515625, + "0.6": 0.4178873697916667, + "0.7": 0.3681640625, + "0.8": 0.304443359375, + "0.9": 0.19327799479166666 + }, + "matched_cos_abs": { + "0.5": 0.7702289819717407, + "0.6": 0.7701990604400635, + "0.7": 0.7701990604400635, + "0.8": 0.7701990604400635, + "0.9": 0.7701990604400635 + }, + "unique_frac_abs": { + "0.5": 0.9999593098958334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7701976412742322, + "0.6": 0.7701990604400635, + "0.7": 0.7701990604400635, + "0.8": 0.7701990604400635, + "0.9": 0.7701990604400635 + }, + "matched_frac_abs": { + "0.5": 0.8655598958333334, + "0.6": 0.8194173177083334, + "0.7": 0.7435709635416666, + "0.8": 0.612548828125, + "0.9": 0.3902994791666667 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.4908197820186615, + "dcs_abs": 0.7686047554016113, + "frac_sign_flipped": 0.5050455729166666, + "alive_source": "stats", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.5": 0.4908459782600403, + "0.6": 0.4908197820186615, + "0.7": 0.4908197820186615, + "0.8": 0.4908197820186615, + "0.9": 0.4908197820186615 + }, + "unique_frac_signed": { + "0.5": 0.9999593098958334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.4908260056860551, + "0.6": 0.4908197820186615, + "0.7": 0.4908197820186615, + "0.8": 0.4908197820186615, + "0.9": 0.4908197820186615 + }, + "matched_frac_signed": { + "0.5": 0.4567057291666667, + "0.6": 0.415283203125, + "0.7": 0.3661295572916667, + "0.8": 0.2996419270833333, + "0.9": 0.191650390625 + }, + "matched_cos_abs": { + "0.5": 0.7686271071434021, + "0.6": 0.7686047554016113, + "0.7": 0.7686047554016113, + "0.8": 0.7686047554016113, + "0.9": 0.7686047554016113 + }, + "unique_frac_abs": { + "0.5": 0.9999593098958334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7685958316263471, + "0.6": 0.7686047554016113, + "0.7": 0.7686047554016113, + "0.8": 0.7686047554016113, + "0.9": 0.7686047554016113 + }, + "matched_frac_abs": { + "0.5": 0.8640950520833334, + "0.6": 0.81689453125, + "0.7": 0.736328125, + "0.8": 0.6095377604166666, + "0.9": 0.390380859375 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 73, + "n_dense_std": 2.6457513110645907, + "l0_mean": 64, + "l0_std": 0.0, + "recon_r2_mean": 0.9257394439691509, + "recon_r2_std": null, + "l1_per_token_mean": 3.3933671366850438, + "l1_per_token_std": 0.005480913978130949, + "babel_mu1_signed_mean": 0.1622351755698522, + "babel_mu1_signed_std": 0.0006198631473248853, + "n_dup_signed_per_threshold": { + "0.5": 0.6666666666666666, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_signed_band": 0.0, + "n_dup_signed_band_std": 0.0, + "n_dup_clusters_signed_per_threshold": { + "0.5": 0.3333333333333333, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_signed_band": 0.0, + "n_dup_clusters_signed_band_std": 0.0, + "largest_dup_signed_per_threshold": { + "0.5": 1.3333333333333333, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_signed_band": 1.0, + "largest_dup_signed_band_std": 0.0, + "dcs_signed_mean": 0.4915749728679657, + "dcs_signed_std": 0.0009632213335710898, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.491574378514795, + "0.6": 0.4915749728679657, + "0.7": 0.4915749728679657, + "0.8": 0.4915749728679657, + "0.9": 0.4915749728679657 + }, + "dedup_dcs_signed_band": 0.4915749728679657, + "dedup_dcs_signed_band_std": 0.0009632213335710898, + "matched_cos_signed_per_threshold": { + "0.5": 0.49158771832784015, + "0.6": 0.4915749728679657, + "0.7": 0.4915749728679657, + "0.8": 0.4915749728679657, + "0.9": 0.4915749728679657 + }, + "matched_cos_signed_band": 0.4915749728679657, + "matched_cos_signed_band_std": 0.0009632213335710898, + "unique_frac_signed_per_threshold": { + "0.5": 0.999972873263889, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_signed_band": 1.0, + "unique_frac_signed_band_std": 0.0, + "matched_frac_signed_per_threshold": { + "0.5": 0.4570041232638889, + "0.6": 0.4154188368055556, + "0.7": 0.3659396701388889, + "0.8": 0.3015950520833333, + "0.9": 0.1924370659722222 + }, + "matched_frac_signed_band": 0.31884765625, + "matched_frac_signed_band_std": 0.0018534107953516968, + "babel_mu1_abs_mean": 0.17155892153580984, + "babel_mu1_abs_std": 0.0008203314925571085, + "n_dup_abs_per_threshold": { + "0.5": 0.6666666666666666, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_abs_band": 0.0, + "n_dup_abs_band_std": 0.0, + "n_dup_clusters_abs_per_threshold": { + "0.5": 0.3333333333333333, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_abs_band": 0.0, + "n_dup_clusters_abs_band_std": 0.0, + "largest_dup_abs_per_threshold": { + "0.5": 1.3333333333333333, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_abs_band": 1.0, + "largest_dup_abs_band_std": 0.0, + "dcs_abs_mean": 0.7690805395444235, + "dcs_abs_std": 0.0009722577019419036, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.7690770918973916, + "0.6": 0.7690805395444235, + "0.7": 0.7690805395444235, + "0.8": 0.7690805395444235, + "0.9": 0.7690805395444235 + }, + "dedup_dcs_abs_band": 0.7690805395444235, + "dedup_dcs_abs_band_std": 0.0009722577019419036, + "matched_cos_abs_per_threshold": { + "0.5": 0.7690979639689127, + "0.6": 0.7690805395444235, + "0.7": 0.7690805395444235, + "0.8": 0.7690805395444235, + "0.9": 0.7690805395444235 + }, + "matched_cos_abs_band": 0.7690805395444235, + "matched_cos_abs_band_std": 0.0009722577019419036, + "unique_frac_abs_per_threshold": { + "0.5": 0.999972873263889, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_abs_band": 1.0, + "unique_frac_abs_band_std": 0.0, + "matched_frac_abs_per_threshold": { + "0.5": 0.8650716145833334, + "0.6": 0.816867404513889, + "0.7": 0.7387152777777777, + "0.8": 0.6105685763888888, + "0.9": 0.38959418402777785 + }, + "matched_frac_abs_band": 0.6389363606770833, + "matched_frac_abs_band_std": 0.002268450498822649, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/config.json b/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..fc536eebc53f7224072b82e212ac5b2c9a4043f6 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 64, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/lock.lock b/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..da5925016372302500e2e770026e44b2db58de45 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 63.94238857421875, + "l0_per_token_0.05": 17.11366591796875, + "l0_per_token_0.1": 5.366815283203125, + "l0_per_token_0.5": 0.372700341796875, + "l0_per_token_1.0": 0.0153291015625, + "l1_per_token": 3.3973256462022587, + "r2_score": 0.9257008624355846, + "l2_loss": 0.04946621021842547 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/config.json b/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ba8c1310f91975954f6b4828f0c9c0dc066d6f63 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 64, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/lock.lock b/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..db452762a72aeb96689bed3030d901c38ca789d9 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 63.9381724609375, + "l0_per_token_0.05": 16.954077392578125, + "l0_per_token_0.1": 5.33339853515625, + "l0_per_token_0.5": 0.37267353515625, + "l0_per_token_1.0": 0.015487841796875, + "l1_per_token": 3.3871114266376185, + "r2_score": 0.925800207561149, + "l2_loss": 0.04940006911603264 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/config.json b/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..72c099534805adbcab623b82c98d6f0a5bd89862 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 64, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/lock.lock b/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..cee7cc2c39cc205abada891111fe2e0c97045ba0 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 63.93735390625, + "l0_per_token_0.05": 17.071048095703127, + "l0_per_token_0.1": 5.363772900390625, + "l0_per_token_0.5": 0.372700146484375, + "l0_per_token_1.0": 0.015348828125, + "l1_per_token": 3.3956643372152553, + "r2_score": 0.925717261910719, + "l2_loss": 0.04945529192366374 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/code_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x4/l0_128/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..63285016cae68be616d2912b5d89c33081e4ea1c --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 128.0, + "recon_r2": 0.9397314587235996, + "l1_per_token": 4.88383873318295 + }, + "seed1": { + "l0": 128.0, + "recon_r2": 0.9398118799965881, + "l1_per_token": 4.881521623246374 + }, + "seed2": { + "l0": 128.0, + "recon_r2": 0.9397466628714934, + "l1_per_token": 4.883710121506085 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.43779266427989083, + "mcc_abs": 0.44402084839385086, + "act_iou": 0.25212282175589434, + "act_iou_n_pairs": 3072, + "input_r2": 0.9641671738177685, + "matched_mcc_signed": { + "0.6": 0.8353503921417601, + "0.7": 0.8547656429645689, + "0.8": 0.8733860935672849, + "0.9": 0.9038527329518758 + }, + "matched_mcc_abs": { + "0.6": 0.8358522790552554, + "0.7": 0.8544736872266971, + "0.8": 0.8737071908922963, + "0.9": 0.9045858238634928 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.44367984669239435, + "mcc_abs": 0.44965580855891285, + "act_iou": 0.2547510053441082, + "act_iou_n_pairs": 3072, + "input_r2": 0.9634377178089526, + "matched_mcc_signed": { + "0.6": 0.8286602086626218, + "0.7": 0.8476904426463725, + "0.8": 0.8703714497982498, + "0.9": 0.9075264874789908 + }, + "matched_mcc_abs": { + "0.6": 0.8306682714015043, + "0.7": 0.8479193630979883, + "0.8": 0.8716540575592054, + "0.9": 0.9066519796793303 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.43581247396650996, + "mcc_abs": 0.44176213821570337, + "act_iou": 0.25138839437290034, + "act_iou_n_pairs": 3072, + "input_r2": 0.9646090148029939, + "matched_mcc_signed": { + "0.6": 0.8275757759456907, + "0.7": 0.8478785928604298, + "0.8": 0.870307298008206, + "0.9": 0.9031727743713549 + }, + "matched_mcc_abs": { + "0.6": 0.8348033309158736, + "0.7": 0.8524999742040517, + "0.8": 0.8742236950323955, + "0.9": 0.9042351561838555 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 128.0, + "l0_std": 0.0, + "recon_r2_mean": 0.9397633338638937, + "recon_r2_std": 4.272396023286045e-05, + "l1_per_token_mean": 4.883023492645136, + "l1_per_token_std": 0.0013022457560155898, + "input_r2_mean": 0.9640713021432382, + "input_r2_std": 0.0005915046031481506, + "mcc_signed_mean": 0.43909499497959836, + "mcc_signed_std": 0.004092180022988721, + "mcc_abs_mean": 0.4451462650561557, + "mcc_abs_std": 0.004065394183798323, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8305287922500243, + "0.7": 0.850111559490457, + "0.8": 0.8713549471245803, + "0.9": 0.9048506649340737 + }, + "matched_mcc_signed_band": 0.8642114909497838, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8337746271242111, + "0.7": 0.8516310081762457, + "0.8": 0.8731949811612991, + "0.9": 0.9051576532422262 + }, + "matched_mcc_abs_band": 0.8659395674259955, + "act_iou_mean": 0.252754073824301, + "act_iou_std": 0.0017679500887532514, + "act_iou_n_pairs": 3072.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/dict_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x4/l0_128/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..d4b2000a5e73e4ffea35f4ba374f6a00444f7548 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 3072, + "d_in": 768, + "l0": 128, + "l0_raw": 128.0, + "recon_r2": 0.9399857276205903, + "l1_per_token": 4.8779511890555876, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 296, + "dense_rate": 0.09635416666666667, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.1294705867767334, + "n_dup_abs": { + "0.5": 4, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.13682880997657776 + }, + "seed1": { + "d_sae": 3072, + "d_in": 768, + "l0": 128, + "l0_raw": 128.0, + "recon_r2": 0.9400871371388, + "l1_per_token": 4.875161542336191, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 295, + "dense_rate": 0.09602864583333333, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.12921933829784393, + "n_dup_abs": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.1371503323316574 + }, + "seed2": { + "d_sae": 3072, + "d_in": 768, + "l0": 128, + "l0_raw": 128.0, + "recon_r2": 0.9400219887369567, + "l1_per_token": 4.877215419157948, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 297, + "dense_rate": 0.0966796875, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.12861792743206024, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.13670885562896729 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.47502341866493225, + "dcs_abs": 0.79034423828125, + "frac_sign_flipped": 0.5172526041666666, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.4750509560108185, + "0.6": 0.47502341866493225, + "0.7": 0.47502341866493225, + "0.8": 0.47502341866493225, + "0.9": 0.47502341866493225 + }, + "unique_frac_signed": { + "0.5": 0.9996744791666666, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.47489631702774204, + "0.6": 0.47502341866493225, + "0.7": 0.47502341866493225, + "0.8": 0.47502341866493225, + "0.9": 0.47502341866493225 + }, + "matched_frac_signed": { + "0.5": 0.4436848958333333, + "0.6": 0.4091796875, + "0.7": 0.3743489583333333, + "0.8": 0.3251953125, + "0.9": 0.22428385416666666 + }, + "matched_cos_abs": { + "0.5": 0.7904996871948242, + "0.6": 0.79034423828125, + "0.7": 0.79034423828125, + "0.8": 0.79034423828125, + "0.9": 0.79034423828125 + }, + "unique_frac_abs": { + "0.5": 0.99951171875, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7901137010194361, + "0.6": 0.79034423828125, + "0.7": 0.79034423828125, + "0.8": 0.79034423828125, + "0.9": 0.79034423828125 + }, + "matched_frac_abs": { + "0.5": 0.8782552083333334, + "0.6": 0.8343098958333334, + "0.7": 0.7727864583333334, + "0.8": 0.6741536458333334, + "0.9": 0.4674479166666667 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.48455366492271423, + "dcs_abs": 0.782008945941925, + "frac_sign_flipped": 0.5048828125, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.48461446166038513, + "0.6": 0.48455366492271423, + "0.7": 0.48455366492271423, + "0.8": 0.48455366492271423, + "0.9": 0.48455366492271423 + }, + "unique_frac_signed": { + "0.5": 0.9998372395833334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.4845355856086826, + "0.6": 0.48455366492271423, + "0.7": 0.48455366492271423, + "0.8": 0.48455366492271423, + "0.9": 0.48455366492271423 + }, + "matched_frac_signed": { + "0.5": 0.4563802083333333, + "0.6": 0.4222005208333333, + "0.7": 0.3873697916666667, + "0.8": 0.3310546875, + "0.9": 0.21354166666666666 + }, + "matched_cos_abs": { + "0.5": 0.7822115421295166, + "0.6": 0.782008945941925, + "0.7": 0.782008945941925, + "0.8": 0.782008945941925, + "0.9": 0.782008945941925 + }, + "unique_frac_abs": { + "0.5": 0.9996744791666666, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7819569159764796, + "0.6": 0.782008945941925, + "0.7": 0.782008945941925, + "0.8": 0.782008945941925, + "0.9": 0.782008945941925 + }, + "matched_frac_abs": { + "0.5": 0.8720703125, + "0.6": 0.8323567708333334, + "0.7": 0.7705078125, + "0.8": 0.658203125, + "0.9": 0.4300130208333333 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.4764353036880493, + "dcs_abs": 0.7941438555717468, + "frac_sign_flipped": 0.5143229166666666, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.4765327572822571, + "0.6": 0.4764353036880493, + "0.7": 0.4764353036880493, + "0.8": 0.4764353036880493, + "0.9": 0.4764353036880493 + }, + "unique_frac_signed": { + "0.5": 0.9998372395833334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.47645519661212654, + "0.6": 0.4764353036880493, + "0.7": 0.4764353036880493, + "0.8": 0.4764353036880493, + "0.9": 0.4764353036880493 + }, + "matched_frac_signed": { + "0.5": 0.4456380208333333, + "0.6": 0.4143880208333333, + "0.7": 0.3766276041666667, + "0.8": 0.3304036458333333, + "0.9": 0.2236328125 + }, + "matched_cos_abs": { + "0.5": 0.794196367263794, + "0.6": 0.7941438555717468, + "0.7": 0.7941438555717468, + "0.8": 0.7941438555717468, + "0.9": 0.7941438555717468 + }, + "unique_frac_abs": { + "0.5": 0.9998372395833334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7940671035321429, + "0.6": 0.7941438555717468, + "0.7": 0.7941438555717468, + "0.8": 0.7941438555717468, + "0.9": 0.7941438555717468 + }, + "matched_frac_abs": { + "0.5": 0.8805338541666666, + "0.6": 0.8450520833333334, + "0.7": 0.7822265625, + "0.8": 0.6829427083333334, + "0.9": 0.4742838541666667 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 296, + "n_dense_std": 1.0, + "l0_mean": 128, + "l0_std": 0.0, + "recon_r2_mean": 0.9400316178321156, + "recon_r2_std": null, + "l1_per_token_mean": 4.876776050183242, + "l1_per_token_std": 0.0014457925857928819, + "babel_mu1_signed_mean": 0.12910261750221252, + "babel_mu1_signed_std": 0.0004381492869037174, + "n_dup_signed_per_threshold": { + "0.5": 1.3333333333333333, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_signed_band": 0.0, + "n_dup_signed_band_std": 0.0, + "n_dup_clusters_signed_per_threshold": { + "0.5": 0.6666666666666666, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_signed_band": 0.0, + "n_dup_clusters_signed_band_std": 0.0, + "largest_dup_signed_per_threshold": { + "0.5": 1.6666666666666667, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_signed_band": 1.0, + "largest_dup_signed_band_std": 0.0, + "dcs_signed_mean": 0.47867079575856525, + "dcs_signed_std": 0.0051433906162663395, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.47862903308285043, + "0.6": 0.47867079575856525, + "0.7": 0.47867079575856525, + "0.8": 0.47867079575856525, + "0.9": 0.47867079575856525 + }, + "dedup_dcs_signed_band": 0.47867079575856525, + "dedup_dcs_signed_band_std": 0.0051433906162663395, + "matched_cos_signed_per_threshold": { + "0.5": 0.4787327249844869, + "0.6": 0.47867079575856525, + "0.7": 0.47867079575856525, + "0.8": 0.47867079575856525, + "0.9": 0.47867079575856525 + }, + "matched_cos_signed_band": 0.47867079575856525, + "matched_cos_signed_band_std": 0.0051433906162663395, + "unique_frac_signed_per_threshold": { + "0.5": 0.9997829861111112, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_signed_band": 1.0, + "unique_frac_signed_band_std": 0.0, + "matched_frac_signed_per_threshold": { + "0.5": 0.4485677083333333, + "0.6": 0.41525607638888884, + "0.7": 0.37944878472222227, + "0.8": 0.3288845486111111, + "0.9": 0.22048611111111108 + }, + "matched_frac_signed_band": 0.33601888020833337, + "matched_frac_signed_band_std": 0.0026532943338055267, + "babel_mu1_abs_mean": 0.13689599931240082, + "babel_mu1_abs_std": 0.00022827883145573092, + "n_dup_abs_per_threshold": { + "0.5": 2.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_abs_band": 0.0, + "n_dup_abs_band_std": 0.0, + "n_dup_clusters_abs_per_threshold": { + "0.5": 1.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_abs_band": 0.0, + "n_dup_clusters_abs_band_std": 0.0, + "largest_dup_abs_per_threshold": { + "0.5": 1.6666666666666667, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_abs_band": 1.0, + "largest_dup_abs_band_std": 0.0, + "dcs_abs_mean": 0.7888323465983073, + "dcs_abs_std": 0.006207122544037521, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.7887125735093529, + "0.6": 0.7888323465983073, + "0.7": 0.7888323465983073, + "0.8": 0.7888323465983073, + "0.9": 0.7888323465983073 + }, + "dedup_dcs_abs_band": 0.7888323465983073, + "dedup_dcs_abs_band_std": 0.006207122544037521, + "matched_cos_abs_per_threshold": { + "0.5": 0.7889691988627116, + "0.6": 0.7888323465983073, + "0.7": 0.7888323465983073, + "0.8": 0.7888323465983073, + "0.9": 0.7888323465983073 + }, + "matched_cos_abs_band": 0.7888323465983073, + "matched_cos_abs_band_std": 0.006207122544037521, + "unique_frac_abs_per_threshold": { + "0.5": 0.9996744791666666, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_abs_band": 1.0, + "unique_frac_abs_band_std": 0.0, + "matched_frac_abs_per_threshold": { + "0.5": 0.876953125, + "0.6": 0.8372395833333334, + "0.7": 0.7751736111111112, + "0.8": 0.6717664930555557, + "0.9": 0.4572482638888889 + }, + "matched_frac_abs_band": 0.6853569878472223, + "matched_frac_abs_band_std": 0.011783655585295346, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/seed0/config.json b/lm/pythia-160m_layer8/isae/x4/l0_128/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..58683470b63be99124282c520391d3e3a47e767a --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/seed0/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 128, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..c1beb2b5ba4ccc63dde8c8a16db1390a537dd0e8 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 127.79749360351562, + "l0_per_token_0.05": 20.107170361328127, + "l0_per_token_0.1": 5.733997021484375, + "l0_per_token_0.5": 0.38299609375, + "l0_per_token_1.0": 0.00788359375, + "l1_per_token": 4.8779511890555876, + "r2_score": 0.9399857276205903, + "l2_loss": 0.03995576141179237 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/seed1/config.json b/lm/pythia-160m_layer8/isae/x4/l0_128/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..aa6d57cd26b5f2599a20f38d50b6f722e672d2fa --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/seed1/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 128, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..21caff4523037508825472f6a994d38dd633c559 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 127.78634033203124, + "l0_per_token_0.05": 20.11730830078125, + "l0_per_token_0.1": 5.772030615234375, + "l0_per_token_0.5": 0.384430712890625, + "l0_per_token_1.0": 0.008452880859375, + "l1_per_token": 4.875161542336191, + "r2_score": 0.9400871371388, + "l2_loss": 0.03988824589666868 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/seed2/config.json b/lm/pythia-160m_layer8/isae/x4/l0_128/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2bd41291b9360c7eb800a0914203b4894f6d83f4 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/seed2/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 128, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..95a1caddf2a59a479df1ac5942ed991b28b7d029 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 127.79403828125, + "l0_per_token_0.05": 20.115390576171876, + "l0_per_token_0.1": 5.7746296875, + "l0_per_token_0.5": 0.38325771484375, + "l0_per_token_1.0": 0.00703388671875, + "l1_per_token": 4.877215419157948, + "r2_score": 0.9400219887369567, + "l2_loss": 0.03993161981252574 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/code_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x4/l0_32/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..f7172394f48bade5b011e8a09e38809af075a06a --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 32.0, + "recon_r2": 0.8887052617232106, + "l1_per_token": 2.5484596555305226 + }, + "seed1": { + "l0": 32.0, + "recon_r2": 0.8887272358629924, + "l1_per_token": 2.5468695358972777 + }, + "seed2": { + "l0": 32.0, + "recon_r2": 0.8887162710620403, + "l1_per_token": 2.550124905518714 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.4444564756105831, + "mcc_abs": 0.4479632348381523, + "act_iou": 0.28680923991804586, + "act_iou_n_pairs": 3072, + "input_r2": 0.9657435367355842, + "matched_mcc_signed": { + "0.6": 0.8652345044765074, + "0.7": 0.8844438418950169, + "0.8": 0.904707944104763, + "0.9": 0.925581108937863 + }, + "matched_mcc_abs": { + "0.6": 0.8697243643894671, + "0.7": 0.8845079348879255, + "0.8": 0.9043201506674756, + "0.9": 0.9248008695412955 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.4546143449047417, + "mcc_abs": 0.4575207263314865, + "act_iou": 0.29842367396848873, + "act_iou_n_pairs": 3072, + "input_r2": 0.9669704208129724, + "matched_mcc_signed": { + "0.6": 0.8653193422188546, + "0.7": 0.8853393822697199, + "0.8": 0.9060162628118401, + "0.9": 0.9250402604027882 + }, + "matched_mcc_abs": { + "0.6": 0.870020896595758, + "0.7": 0.8840669026313165, + "0.8": 0.9046451655701921, + "0.9": 0.9250998820863563 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.4513201964190532, + "mcc_abs": 0.4553980678069802, + "act_iou": 0.29191723280826926, + "act_iou_n_pairs": 3072, + "input_r2": 0.9652669607380323, + "matched_mcc_signed": { + "0.6": 0.8614779406239327, + "0.7": 0.8834000028119136, + "0.8": 0.9002668010382276, + "0.9": 0.9237666513839724 + }, + "matched_mcc_abs": { + "0.6": 0.8690565198635749, + "0.7": 0.8853154178542677, + "0.8": 0.9022817984465199, + "0.9": 0.9253818291499345 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 32.0, + "l0_std": 0.0, + "recon_r2_mean": 0.8887162562160812, + "recon_r2_std": 1.0987077413450392e-05, + "l1_per_token_mean": 2.5484846989821714, + "l1_per_token_std": 0.0016278292981438615, + "input_r2_mean": 0.965993639428863, + "input_r2_std": 0.0008788387648582906, + "mcc_signed_mean": 0.450130338978126, + "mcc_signed_std": 0.005182412343359004, + "mcc_abs_mean": 0.45362734299220636, + "mcc_abs_std": 0.005018765853128715, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8640105957730982, + "0.7": 0.8843944089922168, + "0.8": 0.9036636693182768, + "0.9": 0.924796006908208 + }, + "matched_mcc_signed_band": 0.89421617024795, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8696005936162666, + "0.7": 0.8846300851245031, + "0.8": 0.9037490382280625, + "0.9": 0.9250941935925288 + }, + "matched_mcc_abs_band": 0.8957684776403403, + "act_iou_mean": 0.2923833822316013, + "act_iou_std": 0.005821231917874113, + "act_iou_n_pairs": 3072.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/dict_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x4/l0_32/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..784ebe00c11f720569b5e327bf72e8b76599e215 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 3072, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.8891959164586675, + "l1_per_token": 2.5450888325374144, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 20, + "dense_rate": 0.006510416666666667, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.16527600586414337, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.17755226790905 + }, + "seed1": { + "d_sae": 3072, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.8892593857654404, + "l1_per_token": 2.5433750642904625, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 18, + "dense_rate": 0.005859375, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.16603626310825348, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.17885620892047882 + }, + "seed2": { + "d_sae": 3072, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.8892087636818485, + "l1_per_token": 2.5469234020666973, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 19, + "dense_rate": 0.006184895833333333, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.1649484485387802, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.17736203968524933 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.5153506398200989, + "dcs_abs": 0.8406197428703308, + "frac_sign_flipped": 0.5113932291666666, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.5153506398200989, + "0.6": 0.5153506398200989, + "0.7": 0.5153506398200989, + "0.8": 0.5153506398200989, + "0.9": 0.5153506398200989 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.5153506398200989, + "0.6": 0.5153506398200989, + "0.7": 0.5153506398200989, + "0.8": 0.5153506398200989, + "0.9": 0.5153506398200989 + }, + "matched_frac_signed": { + "0.5": 0.4658203125, + "0.6": 0.4388020833333333, + "0.7": 0.4117838541666667, + "0.8": 0.373046875, + "0.9": 0.314453125 + }, + "matched_cos_abs": { + "0.5": 0.8406197428703308, + "0.6": 0.8406197428703308, + "0.7": 0.8406197428703308, + "0.8": 0.8406197428703308, + "0.9": 0.8406197428703308 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.8406197428703308, + "0.6": 0.8406197428703308, + "0.7": 0.8406197428703308, + "0.8": 0.8406197428703308, + "0.9": 0.8406197428703308 + }, + "matched_frac_abs": { + "0.5": 0.9026692708333334, + "0.6": 0.8759765625, + "0.7": 0.8352864583333334, + "0.8": 0.7620442708333334, + "0.9": 0.6412760416666666 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.5270854830741882, + "dcs_abs": 0.8439741730690002, + "frac_sign_flipped": 0.494140625, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.5270854830741882, + "0.6": 0.5270854830741882, + "0.7": 0.5270854830741882, + "0.8": 0.5270854830741882, + "0.9": 0.5270854830741882 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.5270854830741882, + "0.6": 0.5270854830741882, + "0.7": 0.5270854830741882, + "0.8": 0.5270854830741882, + "0.9": 0.5270854830741882 + }, + "matched_frac_signed": { + "0.5": 0.4817708333333333, + "0.6": 0.4580078125, + "0.7": 0.427734375, + "0.8": 0.3889973958333333, + "0.9": 0.3313802083333333 + }, + "matched_cos_abs": { + "0.5": 0.8439741730690002, + "0.6": 0.8439741730690002, + "0.7": 0.8439741730690002, + "0.8": 0.8439741730690002, + "0.9": 0.8439741730690002 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.8439741730690002, + "0.6": 0.8439741730690002, + "0.7": 0.8439741730690002, + "0.8": 0.8439741730690002, + "0.9": 0.8439741730690002 + }, + "matched_frac_abs": { + "0.5": 0.9059244791666666, + "0.6": 0.8824869791666666, + "0.7": 0.841796875, + "0.8": 0.7646484375, + "0.9": 0.6468098958333334 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.5236314535140991, + "dcs_abs": 0.840806782245636, + "frac_sign_flipped": 0.4921875, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.5236314535140991, + "0.6": 0.5236314535140991, + "0.7": 0.5236314535140991, + "0.8": 0.5236314535140991, + "0.9": 0.5236314535140991 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.5236314535140991, + "0.6": 0.5236314535140991, + "0.7": 0.5236314535140991, + "0.8": 0.5236314535140991, + "0.9": 0.5236314535140991 + }, + "matched_frac_signed": { + "0.5": 0.478515625, + "0.6": 0.4541015625, + "0.7": 0.4228515625, + "0.8": 0.3909505208333333, + "0.9": 0.3196614583333333 + }, + "matched_cos_abs": { + "0.5": 0.840806782245636, + "0.6": 0.840806782245636, + "0.7": 0.840806782245636, + "0.8": 0.840806782245636, + "0.9": 0.840806782245636 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.840806782245636, + "0.6": 0.840806782245636, + "0.7": 0.840806782245636, + "0.8": 0.840806782245636, + "0.9": 0.840806782245636 + }, + "matched_frac_abs": { + "0.5": 0.8990885416666666, + "0.6": 0.876953125, + "0.7": 0.833984375, + "0.8": 0.7698567708333334, + "0.9": 0.6370442708333334 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 19, + "n_dense_std": 1.0, + "l0_mean": 32, + "l0_std": 0.0, + "recon_r2_mean": 0.8892213553019855, + "recon_r2_std": null, + "l1_per_token_mean": 2.545129099631525, + "l1_per_token_std": 0.0017745123001276049, + "babel_mu1_signed_mean": 0.16542023917039236, + "babel_mu1_signed_std": 0.0005580659184748803, + "n_dup_signed_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_signed_band": 0.0, + "n_dup_signed_band_std": 0.0, + "n_dup_clusters_signed_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_signed_band": 0.0, + "n_dup_clusters_signed_band_std": 0.0, + "largest_dup_signed_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_signed_band": 1.0, + "largest_dup_signed_band_std": 0.0, + "dcs_signed_mean": 0.522022525469462, + "dcs_signed_std": 0.006030598945306038, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.522022525469462, + "0.6": 0.522022525469462, + "0.7": 0.522022525469462, + "0.8": 0.522022525469462, + "0.9": 0.522022525469462 + }, + "dedup_dcs_signed_band": 0.522022525469462, + "dedup_dcs_signed_band_std": 0.006030598945306038, + "matched_cos_signed_per_threshold": { + "0.5": 0.522022525469462, + "0.6": 0.522022525469462, + "0.7": 0.522022525469462, + "0.8": 0.522022525469462, + "0.9": 0.522022525469462 + }, + "matched_cos_signed_band": 0.522022525469462, + "matched_cos_signed_band_std": 0.006030598945306038, + "unique_frac_signed_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_signed_band": 1.0, + "unique_frac_signed_band_std": 0.0, + "matched_frac_signed_per_threshold": { + "0.5": 0.4753689236111111, + "0.6": 0.4503038194444444, + "0.7": 0.4207899305555556, + "0.8": 0.38433159722222215, + "0.9": 0.32183159722222215 + }, + "matched_frac_signed_band": 0.39431423611111105, + "matched_frac_signed_band_std": 0.008792201607948321, + "babel_mu1_abs_mean": 0.17792350550492605, + "babel_mu1_abs_std": 0.0008133255438755431, + "n_dup_abs_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_abs_band": 0.0, + "n_dup_abs_band_std": 0.0, + "n_dup_clusters_abs_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_abs_band": 0.0, + "n_dup_clusters_abs_band_std": 0.0, + "largest_dup_abs_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_abs_band": 1.0, + "largest_dup_abs_band_std": 0.0, + "dcs_abs_mean": 0.8418002327283224, + "dcs_abs_std": 0.001885008855575667, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.8418002327283224, + "0.6": 0.8418002327283224, + "0.7": 0.8418002327283224, + "0.8": 0.8418002327283224, + "0.9": 0.8418002327283224 + }, + "dedup_dcs_abs_band": 0.8418002327283224, + "dedup_dcs_abs_band_std": 0.001885008855575667, + "matched_cos_abs_per_threshold": { + "0.5": 0.8418002327283224, + "0.6": 0.8418002327283224, + "0.7": 0.8418002327283224, + "0.8": 0.8418002327283224, + "0.9": 0.8418002327283224 + }, + "matched_cos_abs_band": 0.8418002327283224, + "matched_cos_abs_band_std": 0.001885008855575667, + "unique_frac_abs_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_abs_band": 1.0, + "unique_frac_abs_band_std": 0.0, + "matched_frac_abs_per_threshold": { + "0.5": 0.9025607638888888, + "0.6": 0.8784722222222222, + "0.7": 0.8370225694444445, + "0.8": 0.7655164930555557, + "0.9": 0.6417100694444445 + }, + "matched_frac_abs_band": 0.7806803385416667, + "matched_frac_abs_band_std": 0.0028483072916666375, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/seed0/config.json b/lm/pythia-160m_layer8/isae/x4/l0_32/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7698993f019b0b847dd5844e172c86abe89b8c5d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/seed0/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 32, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..48f7410ef2a0a5be1b24488acc9b44b1ce6ba4ce --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 31.994750341796873, + "l0_per_token_0.05": 17.45073779296875, + "l0_per_token_0.1": 5.491682470703125, + "l0_per_token_0.5": 0.3642595703125, + "l0_per_token_1.0": 0.01743134765625, + "l1_per_token": 2.5450888325374144, + "r2_score": 0.8891959164586675, + "l2_loss": 0.07377014416571924 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/seed1/config.json b/lm/pythia-160m_layer8/isae/x4/l0_32/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..8dd350b779af726863c7ffbafe9691b8d48a8977 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/seed1/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 32, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..5cd16e123c690097f406f8e13e0a2eeb7f335e6b --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 32.0, + "l0_per_token_0.05": 17.42281142578125, + "l0_per_token_0.1": 5.48344384765625, + "l0_per_token_0.5": 0.3638255859375, + "l0_per_token_1.0": 0.0175236328125, + "l1_per_token": 2.5433750642904625, + "r2_score": 0.8892593857654404, + "l2_loss": 0.07372788814264596 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/seed2/config.json b/lm/pythia-160m_layer8/isae/x4/l0_32/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c365ecc852813ac60a03d7d03f4fd43a7ef83f74 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/seed2/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 32, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..df79cb492903d9d75573ef1129d37f9b996f5dae --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 31.984328955078126, + "l0_per_token_0.05": 17.506587353515624, + "l0_per_token_0.1": 5.499421533203125, + "l0_per_token_0.5": 0.36205576171875, + "l0_per_token_1.0": 0.017581982421875, + "l1_per_token": 2.5469234020666973, + "r2_score": 0.8892087636818485, + "l2_loss": 0.07376159085724984 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/code_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x4/l0_64/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..f982d21777f8f77e3bd7aacc0ef7f86ad702aff6 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 64.0, + "recon_r2": 0.9134105396267619, + "l1_per_token": 3.463360365829245 + }, + "seed1": { + "l0": 64.0, + "recon_r2": 0.9134991777541218, + "l1_per_token": 3.456165188826832 + }, + "seed2": { + "l0": 64.0, + "recon_r2": 0.9133829041759145, + "l1_per_token": 3.46281811494398 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.45494008760668486, + "mcc_abs": 0.4593258324864524, + "act_iou": 0.2845232927232943, + "act_iou_n_pairs": 3072, + "input_r2": 0.9663753016215331, + "matched_mcc_signed": { + "0.6": 0.8556822776312625, + "0.7": 0.8729067861099598, + "0.8": 0.8933645219109612, + "0.9": 0.9175901796354916 + }, + "matched_mcc_abs": { + "0.6": 0.8595942219981206, + "0.7": 0.8731291649414145, + "0.8": 0.8936781087863708, + "0.9": 0.91816752829099 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.44781367200038696, + "mcc_abs": 0.45446163818998203, + "act_iou": 0.2815351998133413, + "act_iou_n_pairs": 3072, + "input_r2": 0.9655060402682265, + "matched_mcc_signed": { + "0.6": 0.8541327430379766, + "0.7": 0.8703507535553452, + "0.8": 0.8886445372787208, + "0.9": 0.9139052187168009 + }, + "matched_mcc_abs": { + "0.6": 0.8618285471237187, + "0.7": 0.8737331506135826, + "0.8": 0.8925219312681425, + "0.9": 0.9166896742294411 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.4544960391438783, + "mcc_abs": 0.46028427849793424, + "act_iou": 0.28835043457746096, + "act_iou_n_pairs": 3072, + "input_r2": 0.9661716605231112, + "matched_mcc_signed": { + "0.6": 0.8625915381716477, + "0.7": 0.8787711851725474, + "0.8": 0.8958934861800208, + "0.9": 0.9176228965896458 + }, + "matched_mcc_abs": { + "0.6": 0.8625996135000711, + "0.7": 0.8755940504277866, + "0.8": 0.8929298462636927, + "0.9": 0.9159278858383134 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 64.0, + "l0_std": 0.0, + "recon_r2_mean": 0.9134308738522661, + "recon_r2_std": 6.074534379213239e-05, + "l1_per_token_mean": 3.4607812232000192, + "l1_per_token_std": 0.0040067866185356064, + "input_r2_mean": 0.966017667470957, + "input_r2_std": 0.00045463086143808116, + "mcc_signed_mean": 0.45241659958365005, + "mcc_signed_std": 0.003992430526872485, + "mcc_abs_mean": 0.4580239163914562, + "mcc_abs_std": 0.003122022449912084, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8574688529469623, + "0.7": 0.8740095749459508, + "0.8": 0.892634181789901, + "0.9": 0.916372764980646 + }, + "matched_mcc_signed_band": 0.885121343665865, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8613407942073034, + "0.7": 0.8741521219942611, + "0.8": 0.893043295439402, + "0.9": 0.9169283627862481 + }, + "matched_mcc_abs_band": 0.8863661436068037, + "act_iou_mean": 0.28480297570469887, + "act_iou_std": 0.0034162147400422288, + "act_iou_n_pairs": 3072.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/dict_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x4/l0_64/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..b2392e041e572fc8f9ad462b52ec71490fdd333d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 3072, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.913778658508336, + "l1_per_token": 3.459081720196933, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 84, + "dense_rate": 0.02734375, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.13598908483982086, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.14565010368824005 + }, + "seed1": { + "d_sae": 3072, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.9138858438924534, + "l1_per_token": 3.4520332960895597, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 88, + "dense_rate": 0.028645833333333332, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.13714592158794403, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.14636839926242828 + }, + "seed2": { + "d_sae": 3072, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.9137358658788477, + "l1_per_token": 3.4589531188298426, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 84, + "dense_rate": 0.02734375, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.13649213314056396, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.14590893685817719 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.50791335105896, + "dcs_abs": 0.831659734249115, + "frac_sign_flipped": 0.4986979166666667, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.50791335105896, + "0.6": 0.50791335105896, + "0.7": 0.50791335105896, + "0.8": 0.50791335105896, + "0.9": 0.50791335105896 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.50791335105896, + "0.6": 0.50791335105896, + "0.7": 0.50791335105896, + "0.8": 0.50791335105896, + "0.9": 0.50791335105896 + }, + "matched_frac_signed": { + "0.5": 0.4690755208333333, + "0.6": 0.4456380208333333, + "0.7": 0.419921875, + "0.8": 0.376953125, + "0.9": 0.298828125 + }, + "matched_cos_abs": { + "0.5": 0.831659734249115, + "0.6": 0.831659734249115, + "0.7": 0.831659734249115, + "0.8": 0.831659734249115, + "0.9": 0.831659734249115 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.831659734249115, + "0.6": 0.831659734249115, + "0.7": 0.831659734249115, + "0.8": 0.831659734249115, + "0.9": 0.831659734249115 + }, + "matched_frac_abs": { + "0.5": 0.8968098958333334, + "0.6": 0.8727213541666666, + "0.7": 0.8330078125, + "0.8": 0.7473958333333334, + "0.9": 0.5950520833333334 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.5063727498054504, + "dcs_abs": 0.8348063826560974, + "frac_sign_flipped": 0.5006510416666666, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.5063727498054504, + "0.6": 0.5063727498054504, + "0.7": 0.5063727498054504, + "0.8": 0.5063727498054504, + "0.9": 0.5063727498054504 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.5063727498054504, + "0.6": 0.5063727498054504, + "0.7": 0.5063727498054504, + "0.8": 0.5063727498054504, + "0.9": 0.5063727498054504 + }, + "matched_frac_signed": { + "0.5": 0.4677734375, + "0.6": 0.4420572916666667, + "0.7": 0.4176432291666667, + "0.8": 0.376953125, + "0.9": 0.2942708333333333 + }, + "matched_cos_abs": { + "0.5": 0.8348063826560974, + "0.6": 0.8348063826560974, + "0.7": 0.8348063826560974, + "0.8": 0.8348063826560974, + "0.9": 0.8348063826560974 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.8348063826560974, + "0.6": 0.8348063826560974, + "0.7": 0.8348063826560974, + "0.8": 0.8348063826560974, + "0.9": 0.8348063826560974 + }, + "matched_frac_abs": { + "0.5": 0.9046223958333334, + "0.6": 0.8736979166666666, + "0.7": 0.837890625, + "0.8": 0.7578125, + "0.9": 0.5992838541666666 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.5108785033226013, + "dcs_abs": 0.8329472541809082, + "frac_sign_flipped": 0.4934895833333333, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.5108785033226013, + "0.6": 0.5108785033226013, + "0.7": 0.5108785033226013, + "0.8": 0.5108785033226013, + "0.9": 0.5108785033226013 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.5108785033226013, + "0.6": 0.5108785033226013, + "0.7": 0.5108785033226013, + "0.8": 0.5108785033226013, + "0.9": 0.5108785033226013 + }, + "matched_frac_signed": { + "0.5": 0.4733072916666667, + "0.6": 0.4459635416666667, + "0.7": 0.4186197916666667, + "0.8": 0.3811848958333333, + "0.9": 0.3089192708333333 + }, + "matched_cos_abs": { + "0.5": 0.832947313785553, + "0.6": 0.832947313785553, + "0.7": 0.832947313785553, + "0.8": 0.832947313785553, + "0.9": 0.832947313785553 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.832947313785553, + "0.6": 0.832947313785553, + "0.7": 0.832947313785553, + "0.8": 0.832947313785553, + "0.9": 0.832947313785553 + }, + "matched_frac_abs": { + "0.5": 0.8984375, + "0.6": 0.8714192708333334, + "0.7": 0.8343098958333334, + "0.8": 0.7574869791666666, + "0.9": 0.6097005208333334 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 85, + "n_dense_std": 2.3094010767585034, + "l0_mean": 64, + "l0_std": 0.0, + "recon_r2_mean": 0.913800122759879, + "recon_r2_std": null, + "l1_per_token_mean": 3.4566893783721118, + "l1_per_token_std": 0.0040327989576996095, + "babel_mu1_signed_mean": 0.13654237985610962, + "babel_mu1_signed_std": 0.0005800528982510056, + "n_dup_signed_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_signed_band": 0.0, + "n_dup_signed_band_std": 0.0, + "n_dup_clusters_signed_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_signed_band": 0.0, + "n_dup_clusters_signed_band_std": 0.0, + "largest_dup_signed_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_signed_band": 1.0, + "largest_dup_signed_band_std": 0.0, + "dcs_signed_mean": 0.5083882013956705, + "dcs_signed_std": 0.0022901017053910543, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.5083882013956705, + "0.6": 0.5083882013956705, + "0.7": 0.5083882013956705, + "0.8": 0.5083882013956705, + "0.9": 0.5083882013956705 + }, + "dedup_dcs_signed_band": 0.5083882013956705, + "dedup_dcs_signed_band_std": 0.0022901017053910543, + "matched_cos_signed_per_threshold": { + "0.5": 0.5083882013956705, + "0.6": 0.5083882013956705, + "0.7": 0.5083882013956705, + "0.8": 0.5083882013956705, + "0.9": 0.5083882013956705 + }, + "matched_cos_signed_band": 0.5083882013956705, + "matched_cos_signed_band_std": 0.0022901017053910543, + "unique_frac_signed_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_signed_band": 1.0, + "unique_frac_signed_band_std": 0.0, + "matched_frac_signed_per_threshold": { + "0.5": 0.4700520833333333, + "0.6": 0.4445529513888889, + "0.7": 0.41872829861111116, + "0.8": 0.37836371527777773, + "0.9": 0.3006727430555555 + }, + "matched_frac_signed_band": 0.3855794270833333, + "matched_frac_signed_band_std": 0.0029778929790904847, + "babel_mu1_abs_mean": 0.14597581326961517, + "babel_mu1_abs_std": 0.0003637876767837497, + "n_dup_abs_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_abs_band": 0.0, + "n_dup_abs_band_std": 0.0, + "n_dup_clusters_abs_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_abs_band": 0.0, + "n_dup_clusters_abs_band_std": 0.0, + "largest_dup_abs_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_abs_band": 1.0, + "largest_dup_abs_band_std": 0.0, + "dcs_abs_mean": 0.8331377903620402, + "dcs_abs_std": 0.0015819535633296977, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.8331378102302551, + "0.6": 0.8331378102302551, + "0.7": 0.8331378102302551, + "0.8": 0.8331378102302551, + "0.9": 0.8331378102302551 + }, + "dedup_dcs_abs_band": 0.8331378102302551, + "dedup_dcs_abs_band_std": 0.0015819499742009456, + "matched_cos_abs_per_threshold": { + "0.5": 0.8331378102302551, + "0.6": 0.8331378102302551, + "0.7": 0.8331378102302551, + "0.8": 0.8331378102302551, + "0.9": 0.8331378102302551 + }, + "matched_cos_abs_band": 0.8331378102302551, + "matched_cos_abs_band_std": 0.0015819499742009456, + "unique_frac_abs_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_abs_band": 1.0, + "unique_frac_abs_band_std": 0.0, + "matched_frac_abs_per_threshold": { + "0.5": 0.8999565972222223, + "0.6": 0.8726128472222222, + "0.7": 0.8350694444444445, + "0.8": 0.7542317708333334, + "0.9": 0.6013454861111112 + }, + "matched_frac_abs_band": 0.7658148871527778, + "matched_frac_abs_band_std": 0.0033080162129098508, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/seed0/config.json b/lm/pythia-160m_layer8/isae/x4/l0_64/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..584f6af16d8082cd6a6b59a6ab852e3e73b8bac5 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/seed0/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 64, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..d1489e681635d0d958d37574e8302f724d6eaed0 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 63.954517724609374, + "l0_per_token_0.05": 17.72366630859375, + "l0_per_token_0.1": 5.464570068359375, + "l0_per_token_0.5": 0.368371826171875, + "l0_per_token_1.0": 0.01667099609375, + "l1_per_token": 3.459081720196933, + "r2_score": 0.913778658508336, + "l2_loss": 0.05740366770534342 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/seed1/config.json b/lm/pythia-160m_layer8/isae/x4/l0_64/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..aa6ea96215736297f15273926f2155d9e50197ab --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/seed1/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 64, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..b9ee1fb9ce9f614cefe96b90ceb2a168813b07c9 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 63.9433685546875, + "l0_per_token_0.05": 17.64664658203125, + "l0_per_token_0.1": 5.44191083984375, + "l0_per_token_0.5": 0.371726513671875, + "l0_per_token_1.0": 0.01607138671875, + "l1_per_token": 3.4520332960895597, + "r2_score": 0.9138858438924534, + "l2_loss": 0.05733230678626814 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/seed2/config.json b/lm/pythia-160m_layer8/isae/x4/l0_64/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d896eb6ecc74aff8c480e9b1efebbc1e969adec3 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/seed2/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 64, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..331405247366adbe2987d773a58bc82b724df79d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 63.95089765625, + "l0_per_token_0.05": 17.7122248046875, + "l0_per_token_0.1": 5.4809666015625, + "l0_per_token_0.5": 0.37204736328125, + "l0_per_token_1.0": 0.016063134765625, + "l1_per_token": 3.4589531188298426, + "r2_score": 0.9137358658788477, + "l2_loss": 0.05743215779655381 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/code_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x8/l0_128/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..b066285d5eade10cecfc816a6cdd9734e4379a95 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 128.0, + "recon_r2": 0.9435827003863506, + "l1_per_token": 4.796385798898775 + }, + "seed1": { + "l0": 128.0, + "recon_r2": 0.943620818431546, + "l1_per_token": 4.79343959498753 + }, + "seed2": { + "l0": 128.0, + "recon_r2": 0.9435700015802021, + "l1_per_token": 4.803937323151347 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.43136577480490823, + "mcc_abs": 0.4371340212889548, + "act_iou": 0.23064973674930866, + "act_iou_n_pairs": 6144, + "input_r2": 0.9660117879793234, + "matched_mcc_signed": { + "0.6": 0.8008452241713196, + "0.7": 0.8250586694752499, + "0.8": 0.8513189565161097, + "0.9": 0.8938140633998646 + }, + "matched_mcc_abs": { + "0.6": 0.8058759562219018, + "0.7": 0.8260118206655089, + "0.8": 0.852115964420852, + "0.9": 0.8934798297128265 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.44467995544036176, + "mcc_abs": 0.45164562723480545, + "act_iou": 0.24486720281915098, + "act_iou_n_pairs": 6144, + "input_r2": 0.9669695725317103, + "matched_mcc_signed": { + "0.6": 0.8069387692048594, + "0.7": 0.8302211035131151, + "0.8": 0.8572575483051125, + "0.9": 0.8920024574108728 + }, + "matched_mcc_abs": { + "0.6": 0.8077421498170969, + "0.7": 0.8287781293606672, + "0.8": 0.8553048106683188, + "0.9": 0.8920599703185208 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.4311250452268814, + "mcc_abs": 0.4381453388408964, + "act_iou": 0.23157376230097806, + "act_iou_n_pairs": 6144, + "input_r2": 0.9655403442771402, + "matched_mcc_signed": { + "0.6": 0.7999149909671546, + "0.7": 0.8267033131430023, + "0.8": 0.8498767846130606, + "0.9": 0.8932387440816385 + }, + "matched_mcc_abs": { + "0.6": 0.8024159847601787, + "0.7": 0.8264177164745906, + "0.8": 0.8526879000714881, + "0.9": 0.8929615616757388 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 128.0, + "l0_std": 0.0, + "recon_r2_mean": 0.9435911734660328, + "recon_r2_std": 2.6446793843557706e-05, + "l1_per_token_mean": 4.797920905679217, + "l1_per_token_std": 0.005414608828980231, + "input_r2_mean": 0.9661739015960581, + "input_r2_std": 0.0007282746524927805, + "mcc_signed_mean": 0.43572359182405046, + "mcc_signed_std": 0.007757372274367326, + "mcc_abs_mean": 0.4423083291215522, + "mcc_abs_std": 0.008102131993465553, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8025663281144446, + "0.7": 0.8273276953771225, + "0.8": 0.852817763144761, + "0.9": 0.8930184216307918 + }, + "matched_mcc_signed_band": 0.84393255206678, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8053446969330591, + "0.7": 0.8270692221669221, + "0.8": 0.8533695583868863, + "0.9": 0.8928337872356954 + }, + "matched_mcc_abs_band": 0.8446543161806407, + "act_iou_mean": 0.23569690062314588, + "act_iou_std": 0.00795514220989016, + "act_iou_n_pairs": 6144.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/dict_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x8/l0_128/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..06b7ee9dd08302e7b582f386ab17b6963a09dae3 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 6144, + "d_in": 768, + "l0": 128, + "l0_raw": 128.0, + "recon_r2": 0.9438320465887855, + "l1_per_token": 4.79040432951138, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 251, + "dense_rate": 0.040852864583333336, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 3, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 3, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.1429690718650818, + "n_dup_abs": { + "0.5": 5, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 3, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.15059161186218262 + }, + "seed1": { + "d_sae": 6144, + "d_in": 768, + "l0": 128, + "l0_raw": 128.0, + "recon_r2": 0.943894948847332, + "l1_per_token": 4.787156959382066, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 261, + "dense_rate": 0.04248046875, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.14347828924655914, + "n_dup_abs": { + "0.5": 4, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.1509147584438324 + }, + "seed2": { + "d_sae": 6144, + "d_in": 768, + "l0": 128, + "l0_raw": 128.0, + "recon_r2": 0.9438388234176914, + "l1_per_token": 4.797608940777183, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 257, + "dense_rate": 0.041829427083333336, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.14312686026096344, + "n_dup_abs": { + "0.5": 3, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 1, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 3, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.15037505328655243 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.4773045480251312, + "dcs_abs": 0.7554041743278503, + "frac_sign_flipped": 0.5006510416666666, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.4772987365722656, + "0.6": 0.4773045480251312, + "0.7": 0.4773045480251312, + "0.8": 0.4773045480251312, + "0.9": 0.4773045480251312 + }, + "unique_frac_signed": { + "0.5": 0.999755859375, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.47718220856040716, + "0.6": 0.4773045480251312, + "0.7": 0.4773045480251312, + "0.8": 0.4773045480251312, + "0.9": 0.4773045480251312 + }, + "matched_frac_signed": { + "0.5": 0.45263671875, + "0.6": 0.4083658854166667, + "0.7": 0.3544921875, + "0.8": 0.2872721354166667, + "0.9": 0.15755208333333334 + }, + "matched_cos_abs": { + "0.5": 0.7554783821105957, + "0.6": 0.7554041743278503, + "0.7": 0.7554041743278503, + "0.8": 0.7554041743278503, + "0.9": 0.7554041743278503 + }, + "unique_frac_abs": { + "0.5": 0.9995930989583334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7551709771699583, + "0.6": 0.7554041743278503, + "0.7": 0.7554041743278503, + "0.8": 0.7554041743278503, + "0.9": 0.7554041743278503 + }, + "matched_frac_abs": { + "0.5": 0.85693359375, + "0.6": 0.80517578125, + "0.7": 0.720703125, + "0.8": 0.5862630208333334, + "0.9": 0.32666015625 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.4943978488445282, + "dcs_abs": 0.7631834149360657, + "frac_sign_flipped": 0.47802734375, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.49441003799438477, + "0.6": 0.4943978488445282, + "0.7": 0.4943978488445282, + "0.8": 0.4943978488445282, + "0.9": 0.4943978488445282 + }, + "unique_frac_signed": { + "0.5": 0.999755859375, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.49428933241870254, + "0.6": 0.4943978488445282, + "0.7": 0.4943978488445282, + "0.8": 0.4943978488445282, + "0.9": 0.4943978488445282 + }, + "matched_frac_signed": { + "0.5": 0.4742838541666667, + "0.6": 0.4348958333333333, + "0.7": 0.3831380208333333, + "0.8": 0.31005859375, + "0.9": 0.1875 + }, + "matched_cos_abs": { + "0.5": 0.7632564902305603, + "0.6": 0.7631834149360657, + "0.7": 0.7631834149360657, + "0.8": 0.7631834149360657, + "0.9": 0.7631834149360657 + }, + "unique_frac_abs": { + "0.5": 0.9995930989583334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7629459203696267, + "0.6": 0.7631834149360657, + "0.7": 0.7631834149360657, + "0.8": 0.7631834149360657, + "0.9": 0.7631834149360657 + }, + "matched_frac_abs": { + "0.5": 0.8673502604166666, + "0.6": 0.81689453125, + "0.7": 0.7322591145833334, + "0.8": 0.5960286458333334, + "0.9": 0.3548177083333333 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.4792305529117584, + "dcs_abs": 0.7538299560546875, + "frac_sign_flipped": 0.5029296875, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.479216992855072, + "0.6": 0.4792305529117584, + "0.7": 0.4792305529117584, + "0.8": 0.4792305529117584, + "0.9": 0.4792305529117584 + }, + "unique_frac_signed": { + "0.5": 0.9998372395833334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.4791389952976412, + "0.6": 0.4792305529117584, + "0.7": 0.4792305529117584, + "0.8": 0.4792305529117584, + "0.9": 0.4792305529117584 + }, + "matched_frac_signed": { + "0.5": 0.4544270833333333, + "0.6": 0.4109700520833333, + "0.7": 0.3564453125, + "0.8": 0.2892252604166667, + "0.9": 0.16048177083333334 + }, + "matched_cos_abs": { + "0.5": 0.7538555860519409, + "0.6": 0.7538298964500427, + "0.7": 0.7538298964500427, + "0.8": 0.7538298964500427, + "0.9": 0.7538298964500427 + }, + "unique_frac_abs": { + "0.5": 0.9996744791666666, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7536101903533563, + "0.6": 0.7538298964500427, + "0.7": 0.7538298964500427, + "0.8": 0.7538298964500427, + "0.9": 0.7538298964500427 + }, + "matched_frac_abs": { + "0.5": 0.8600260416666666, + "0.6": 0.8094075520833334, + "0.7": 0.7120768229166666, + "0.8": 0.57421875, + "0.9": 0.3225911458333333 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 256, + "n_dense_std": 5.033222956847167, + "l0_mean": 128, + "l0_std": 0.0, + "recon_r2_mean": 0.9438552729512697, + "recon_r2_std": null, + "l1_per_token_mean": 4.791723409890209, + "l1_per_token_std": 0.005349387965570582, + "babel_mu1_signed_mean": 0.14319140712420145, + "babel_mu1_signed_std": 0.00026067279981820926, + "n_dup_signed_per_threshold": { + "0.5": 2.3333333333333335, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_signed_band": 0.0, + "n_dup_signed_band_std": 0.0, + "n_dup_clusters_signed_per_threshold": { + "0.5": 1.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_signed_band": 0.0, + "n_dup_clusters_signed_band_std": 0.0, + "largest_dup_signed_per_threshold": { + "0.5": 2.3333333333333335, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_signed_band": 1.0, + "largest_dup_signed_band_std": 0.0, + "dcs_signed_mean": 0.48364431659380597, + "dcs_signed_std": 0.00936248981875658, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.4835368454255837, + "0.6": 0.48364431659380597, + "0.7": 0.48364431659380597, + "0.8": 0.48364431659380597, + "0.9": 0.48364431659380597 + }, + "dedup_dcs_signed_band": 0.48364431659380597, + "dedup_dcs_signed_band_std": 0.00936248981875658, + "matched_cos_signed_per_threshold": { + "0.5": 0.48364192247390747, + "0.6": 0.48364431659380597, + "0.7": 0.48364431659380597, + "0.8": 0.48364431659380597, + "0.9": 0.48364431659380597 + }, + "matched_cos_signed_band": 0.48364431659380597, + "matched_cos_signed_band_std": 0.00936248981875658, + "unique_frac_signed_per_threshold": { + "0.5": 0.9997829861111112, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_signed_band": 1.0, + "unique_frac_signed_band_std": 0.0, + "matched_frac_signed_per_threshold": { + "0.5": 0.46044921875, + "0.6": 0.4180772569444444, + "0.7": 0.36469184027777773, + "0.8": 0.2955186631944445, + "0.9": 0.16851128472222224 + }, + "matched_frac_signed_band": 0.3116997612847222, + "matched_frac_signed_band_std": 0.014940879515147654, + "babel_mu1_abs_mean": 0.1506271411975225, + "babel_mu1_abs_std": 0.0002716011127571428, + "n_dup_abs_per_threshold": { + "0.5": 4.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_abs_band": 0.0, + "n_dup_abs_band_std": 0.0, + "n_dup_clusters_abs_per_threshold": { + "0.5": 1.6666666666666667, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_abs_band": 0.0, + "n_dup_clusters_abs_band_std": 0.0, + "largest_dup_abs_per_threshold": { + "0.5": 2.6666666666666665, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_abs_band": 1.0, + "largest_dup_abs_band_std": 0.0, + "dcs_abs_mean": 0.7574725151062012, + "dcs_abs_std": 0.005008025902783426, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.7572423626309804, + "0.6": 0.7574724952379862, + "0.7": 0.7574724952379862, + "0.8": 0.7574724952379862, + "0.9": 0.7574724952379862 + }, + "dedup_dcs_abs_band": 0.7574724952379862, + "dedup_dcs_abs_band_std": 0.005008047579403807, + "matched_cos_abs_per_threshold": { + "0.5": 0.757530152797699, + "0.6": 0.7574724952379862, + "0.7": 0.7574724952379862, + "0.8": 0.7574724952379862, + "0.9": 0.7574724952379862 + }, + "matched_cos_abs_band": 0.7574724952379862, + "matched_cos_abs_band_std": 0.005008047579403807, + "unique_frac_abs_per_threshold": { + "0.5": 0.9996202256944445, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_abs_band": 1.0, + "unique_frac_abs_band_std": 0.0, + "matched_frac_abs_per_threshold": { + "0.5": 0.8614366319444443, + "0.6": 0.8104926215277778, + "0.7": 0.7216796875, + "0.8": 0.5855034722222222, + "0.9": 0.33468967013888884 + }, + "matched_frac_abs_band": 0.6130913628472222, + "matched_frac_abs_band_std": 0.010627000574662501, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/seed0/config.json b/lm/pythia-160m_layer8/isae/x8/l0_128/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..40df04c960386578c223597af1001d2e978efd94 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/seed0/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 128, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..cbdc4b0dd8413a3c873cc7cdc56b6bc5237bba61 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 127.77280654296875, + "l0_per_token_0.05": 19.764609521484374, + "l0_per_token_0.1": 5.635467041015625, + "l0_per_token_0.5": 0.38215810546875, + "l0_per_token_1.0": 0.004296875, + "l1_per_token": 4.79040432951138, + "r2_score": 0.9438320465887855, + "l2_loss": 0.03739499383245264 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/seed1/config.json b/lm/pythia-160m_layer8/isae/x8/l0_128/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2577533fca09be994e11f088889f54105f929fcf --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/seed1/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 128, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..cd89dfd40fe72bd3c40377883b9b919e5fdf60e1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 127.76108227539062, + "l0_per_token_0.05": 19.7911525390625, + "l0_per_token_0.1": 5.603636279296875, + "l0_per_token_0.5": 0.381538427734375, + "l0_per_token_1.0": 0.0042982421875, + "l1_per_token": 4.787156959382066, + "r2_score": 0.943894948847332, + "l2_loss": 0.037353115333637166 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/seed2/config.json b/lm/pythia-160m_layer8/isae/x8/l0_128/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a1db69203b94d5ad580031e9ccd7364464b046c4 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/seed2/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 128, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..ec505cdc58448fa73baf1125bc51395df35042dd --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 127.76630419921875, + "l0_per_token_0.05": 19.9138482421875, + "l0_per_token_0.1": 5.6105720703125, + "l0_per_token_0.5": 0.3826185546875, + "l0_per_token_1.0": 0.0049017578125, + "l1_per_token": 4.797608940777183, + "r2_score": 0.9438388234176914, + "l2_loss": 0.03739048201637688 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/code_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x8/l0_32/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..17ccd6e95b5e5c6dc656b584be0f0387f217a21b --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 32.0, + "recon_r2": 0.8973101328596609, + "l1_per_token": 2.5410600410908786 + }, + "seed1": { + "l0": 32.0, + "recon_r2": 0.8974326482136465, + "l1_per_token": 2.53789460054751 + }, + "seed2": { + "l0": 32.0, + "recon_r2": 0.8973347151432118, + "l1_per_token": 2.5404450983888407 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.42369598825137533, + "mcc_abs": 0.4281294963078266, + "act_iou": 0.261239932391585, + "act_iou_n_pairs": 6144, + "input_r2": 0.9670179742222804, + "matched_mcc_signed": { + "0.6": 0.8400973293978742, + "0.7": 0.86393766327794, + "0.8": 0.8858785226170408, + "0.9": 0.9113260891423144 + }, + "matched_mcc_abs": { + "0.6": 0.8459469299665403, + "0.7": 0.8641466913783689, + "0.8": 0.8862206039660105, + "0.9": 0.9124725745463079 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.42683553391231593, + "mcc_abs": 0.4319043079041798, + "act_iou": 0.2656108212496205, + "act_iou_n_pairs": 6144, + "input_r2": 0.9674304030177286, + "matched_mcc_signed": { + "0.6": 0.8426028467385466, + "0.7": 0.8662727367848283, + "0.8": 0.8901598012450056, + "0.9": 0.9169510620257834 + }, + "matched_mcc_abs": { + "0.6": 0.8473420975273047, + "0.7": 0.8644533911048156, + "0.8": 0.8865754057639771, + "0.9": 0.9136272729222633 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.43084131420683164, + "mcc_abs": 0.43479736810176334, + "act_iou": 0.26791503669742583, + "act_iou_n_pairs": 6144, + "input_r2": 0.9678751502410126, + "matched_mcc_signed": { + "0.6": 0.8425477017163224, + "0.7": 0.8654813181999649, + "0.8": 0.8858683584656238, + "0.9": 0.9139258368478602 + }, + "matched_mcc_abs": { + "0.6": 0.8479031574224097, + "0.7": 0.8669050909508147, + "0.8": 0.887624082015223, + "0.9": 0.9140388560696707 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 32.0, + "l0_std": 0.0, + "recon_r2_mean": 0.8973591654055064, + "recon_r2_std": 6.481407628951909e-05, + "l1_per_token_mean": 2.5397999133424096, + "l1_per_token_std": 0.0016784520447435728, + "input_r2_mean": 0.9674411758270072, + "input_r2_std": 0.00042868954015369396, + "mcc_signed_mean": 0.42712427879017434, + "mcc_signed_std": 0.003581403489666652, + "mcc_abs_mean": 0.43161039077125657, + "mcc_abs_std": 0.003343638590796506, + "matched_mcc_signed_per_threshold": { + "0.6": 0.841749292617581, + "0.7": 0.8652305727542444, + "0.8": 0.8873022274425567, + "0.9": 0.9140676626719859 + }, + "matched_mcc_signed_band": 0.877087438871592, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8470640616387516, + "0.7": 0.8651683911446663, + "0.8": 0.8868066972484036, + "0.9": 0.9133795678460807 + }, + "matched_mcc_abs_band": 0.8781046794694756, + "act_iou_mean": 0.2649219301128771, + "act_iou_std": 0.0033904546335523237, + "act_iou_n_pairs": 6144.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/dict_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x8/l0_32/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..741de1d61b7946dd7e89f002311415fceba2c582 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 6144, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.8978087064002357, + "l1_per_token": 2.5377569612014015, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 18, + "dense_rate": 0.0029296875, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.17992663383483887, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.19266708195209503 + }, + "seed1": { + "d_sae": 6144, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.8979187540458593, + "l1_per_token": 2.534355800504567, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 17, + "dense_rate": 0.0027669270833333335, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.1810796707868576, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.19324398040771484 + }, + "seed2": { + "d_sae": 6144, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.897787812974684, + "l1_per_token": 2.5369049473301186, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 17, + "dense_rate": 0.0027669270833333335, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.17963509261608124, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.19179224967956543 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.5110837817192078, + "dcs_abs": 0.8168789744377136, + "frac_sign_flipped": 0.5130208333333334, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.5110837817192078, + "0.6": 0.5110837817192078, + "0.7": 0.5110837817192078, + "0.8": 0.5110837817192078, + "0.9": 0.5110837817192078 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.5110837817192078, + "0.6": 0.5110837817192078, + "0.7": 0.5110837817192078, + "0.8": 0.5110837817192078, + "0.9": 0.5110837817192078 + }, + "matched_frac_signed": { + "0.5": 0.4573567708333333, + "0.6": 0.4288736979166667, + "0.7": 0.39306640625, + "0.8": 0.3517252604166667, + "0.9": 0.2755533854166667 + }, + "matched_cos_abs": { + "0.5": 0.8168789744377136, + "0.6": 0.8168789744377136, + "0.7": 0.8168789744377136, + "0.8": 0.8168789744377136, + "0.9": 0.8168789744377136 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.8168789744377136, + "0.6": 0.8168789744377136, + "0.7": 0.8168789744377136, + "0.8": 0.8168789744377136, + "0.9": 0.8168789744377136 + }, + "matched_frac_abs": { + "0.5": 0.8857421875, + "0.6": 0.85595703125, + "0.7": 0.8038736979166666, + "0.8": 0.7169596354166666, + "0.9": 0.5613606770833334 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.5123879909515381, + "dcs_abs": 0.8174794316291809, + "frac_sign_flipped": 0.5035807291666666, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.5123879909515381, + "0.6": 0.5123879909515381, + "0.7": 0.5123879909515381, + "0.8": 0.5123879909515381, + "0.9": 0.5123879909515381 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.5123879909515381, + "0.6": 0.5123879909515381, + "0.7": 0.5123879909515381, + "0.8": 0.5123879909515381, + "0.9": 0.5123879909515381 + }, + "matched_frac_signed": { + "0.5": 0.4607747395833333, + "0.6": 0.43212890625, + "0.7": 0.3971354166666667, + "0.8": 0.3505859375, + "0.9": 0.2799479166666667 + }, + "matched_cos_abs": { + "0.5": 0.8174794316291809, + "0.6": 0.8174794316291809, + "0.7": 0.8174794316291809, + "0.8": 0.8174794316291809, + "0.9": 0.8174794316291809 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.8174794316291809, + "0.6": 0.8174794316291809, + "0.7": 0.8174794316291809, + "0.8": 0.8174794316291809, + "0.9": 0.8174794316291809 + }, + "matched_frac_abs": { + "0.5": 0.88525390625, + "0.6": 0.8541666666666666, + "0.7": 0.80615234375, + "0.8": 0.7177734375, + "0.9": 0.56787109375 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.518543541431427, + "dcs_abs": 0.8186048865318298, + "frac_sign_flipped": 0.5011393229166666, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.518543541431427, + "0.6": 0.518543541431427, + "0.7": 0.518543541431427, + "0.8": 0.518543541431427, + "0.9": 0.518543541431427 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.518543541431427, + "0.6": 0.518543541431427, + "0.7": 0.518543541431427, + "0.8": 0.518543541431427, + "0.9": 0.518543541431427 + }, + "matched_frac_signed": { + "0.5": 0.4661458333333333, + "0.6": 0.4401041666666667, + "0.7": 0.4041341145833333, + "0.8": 0.36376953125, + "0.9": 0.2877604166666667 + }, + "matched_cos_abs": { + "0.5": 0.8186049461364746, + "0.6": 0.8186049461364746, + "0.7": 0.8186049461364746, + "0.8": 0.8186049461364746, + "0.9": 0.8186049461364746 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.8186049461364746, + "0.6": 0.8186049461364746, + "0.7": 0.8186049461364746, + "0.8": 0.8186049461364746, + "0.9": 0.8186049461364746 + }, + "matched_frac_abs": { + "0.5": 0.8849283854166666, + "0.6": 0.85595703125, + "0.7": 0.8035481770833334, + "0.8": 0.7216796875, + "0.9": 0.5719401041666666 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 17, + "n_dense_std": 0.5773502691896258, + "l0_mean": 32, + "l0_std": 0.0, + "recon_r2_mean": 0.897838424473593, + "recon_r2_std": null, + "l1_per_token_mean": 2.5363392363453623, + "l1_per_token_std": 0.0017697451798851034, + "babel_mu1_signed_mean": 0.18021379907925925, + "babel_mu1_signed_std": 0.0007639040719050768, + "n_dup_signed_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_signed_band": 0.0, + "n_dup_signed_band_std": 0.0, + "n_dup_clusters_signed_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_signed_band": 0.0, + "n_dup_clusters_signed_band_std": 0.0, + "largest_dup_signed_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_signed_band": 1.0, + "largest_dup_signed_band_std": 0.0, + "dcs_signed_mean": 0.5140051047007242, + "dcs_signed_std": 0.003984130570070163, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.5140051047007242, + "0.6": 0.5140051047007242, + "0.7": 0.5140051047007242, + "0.8": 0.5140051047007242, + "0.9": 0.5140051047007242 + }, + "dedup_dcs_signed_band": 0.5140051047007242, + "dedup_dcs_signed_band_std": 0.003984130570070163, + "matched_cos_signed_per_threshold": { + "0.5": 0.5140051047007242, + "0.6": 0.5140051047007242, + "0.7": 0.5140051047007242, + "0.8": 0.5140051047007242, + "0.9": 0.5140051047007242 + }, + "matched_cos_signed_band": 0.5140051047007242, + "matched_cos_signed_band_std": 0.003984130570070163, + "unique_frac_signed_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_signed_band": 1.0, + "unique_frac_signed_band_std": 0.0, + "matched_frac_signed_per_threshold": { + "0.5": 0.46142578125, + "0.6": 0.4337022569444445, + "0.7": 0.3981119791666667, + "0.8": 0.3553602430555556, + "0.9": 0.2810872395833333 + }, + "matched_frac_signed_band": 0.3670654296875, + "matched_frac_signed_band_std": 0.00610039525271484, + "babel_mu1_abs_mean": 0.19256777067979178, + "babel_mu1_abs_std": 0.0007309429344129244, + "n_dup_abs_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_abs_band": 0.0, + "n_dup_abs_band_std": 0.0, + "n_dup_clusters_abs_per_threshold": { + "0.5": 0.0, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_abs_band": 0.0, + "n_dup_clusters_abs_band_std": 0.0, + "largest_dup_abs_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_abs_band": 1.0, + "largest_dup_abs_band_std": 0.0, + "dcs_abs_mean": 0.8176544308662415, + "dcs_abs_std": 0.0008761630492568043, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.8176544507344564, + "0.6": 0.8176544507344564, + "0.7": 0.8176544507344564, + "0.8": 0.8176544507344564, + "0.9": 0.8176544507344564 + }, + "dedup_dcs_abs_band": 0.8176544507344564, + "dedup_dcs_abs_band_std": 0.0008761953786910436, + "matched_cos_abs_per_threshold": { + "0.5": 0.8176544507344564, + "0.6": 0.8176544507344564, + "0.7": 0.8176544507344564, + "0.8": 0.8176544507344564, + "0.9": 0.8176544507344564 + }, + "matched_cos_abs_band": 0.8176544507344564, + "matched_cos_abs_band_std": 0.0008761953786910436, + "unique_frac_abs_per_threshold": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_abs_band": 1.0, + "unique_frac_abs_band_std": 0.0, + "matched_frac_abs_per_threshold": { + "0.5": 0.8853081597222222, + "0.6": 0.8553602430555555, + "0.7": 0.8045247395833334, + "0.8": 0.7188042534722222, + "0.9": 0.5670572916666666 + }, + "matched_frac_abs_band": 0.7364366319444444, + "matched_frac_abs_band_std": 0.0018723344104532949, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/seed0/config.json b/lm/pythia-160m_layer8/isae/x8/l0_32/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4f6f74c4605a9dd3c941492c25517f92ab351c74 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/seed0/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 32, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..0c96e3508a42a238a557038b05a116420b8e01ce --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 31.988375244140624, + "l0_per_token_0.05": 17.16978232421875, + "l0_per_token_0.1": 5.46517744140625, + "l0_per_token_0.5": 0.361009521484375, + "l0_per_token_1.0": 0.01786484375, + "l1_per_token": 2.5377569612014015, + "r2_score": 0.8978087064002357, + "l2_loss": 0.0680359984975089 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/seed1/config.json b/lm/pythia-160m_layer8/isae/x8/l0_32/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca107f7c169ca904540ec33a7211206af747fd8e --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/seed1/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 32, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..4a70791dcf1af568e55a0b4816f2f9083c8cafbc --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 31.985654541015624, + "l0_per_token_0.05": 17.1233330078125, + "l0_per_token_0.1": 5.453684912109375, + "l0_per_token_0.5": 0.360623583984375, + "l0_per_token_1.0": 0.018045068359375, + "l1_per_token": 2.534355800504567, + "r2_score": 0.8979187540458593, + "l2_loss": 0.067962731967763 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/seed2/config.json b/lm/pythia-160m_layer8/isae/x8/l0_32/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c247844b1ed1ce682ed350650507bce658425e5e --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/seed2/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 32, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..279f78a84ce95e28d2c6534e2a2c9eff60cc7fc4 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 31.988471875, + "l0_per_token_0.05": 17.1438287109375, + "l0_per_token_0.1": 5.467050146484375, + "l0_per_token_0.5": 0.360713525390625, + "l0_per_token_1.0": 0.018018603515625, + "l1_per_token": 2.5369049473301186, + "r2_score": 0.897787812974684, + "l2_loss": 0.06804990873407965 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/code_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x8/l0_64/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..ec4bacf891afb1ab2b5014d7b65f0c7c22dde62a --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 64.0, + "recon_r2": 0.9198554227247986, + "l1_per_token": 3.4248076125822653 + }, + "seed1": { + "l0": 64.0, + "recon_r2": 0.9199484637562547, + "l1_per_token": 3.4193746709040926 + }, + "seed2": { + "l0": 64.0, + "recon_r2": 0.9197701261653906, + "l1_per_token": 3.4222683664186198 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.4340457755637222, + "mcc_abs": 0.4404282018411065, + "act_iou": 0.25500859143318405, + "act_iou_n_pairs": 6144, + "input_r2": 0.9671915647590194, + "matched_mcc_signed": { + "0.6": 0.8302428637488269, + "0.7": 0.8523333365581158, + "0.8": 0.8740927164844089, + "0.9": 0.902958495849988 + }, + "matched_mcc_abs": { + "0.6": 0.8366604871684582, + "0.7": 0.8546001594509688, + "0.8": 0.8756346974613898, + "0.9": 0.9033458230566237 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.44082727540262723, + "mcc_abs": 0.44597719873355146, + "act_iou": 0.26069788590223797, + "act_iou_n_pairs": 6144, + "input_r2": 0.9670786747449265, + "matched_mcc_signed": { + "0.6": 0.8315493402949424, + "0.7": 0.852747794204897, + "0.8": 0.8758670726080328, + "0.9": 0.9043921324910993 + }, + "matched_mcc_abs": { + "0.6": 0.8362325461416982, + "0.7": 0.8536087441035809, + "0.8": 0.8763098426117275, + "0.9": 0.903297359321382 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.4405892704686736, + "mcc_abs": 0.4475074969770257, + "act_iou": 0.2604638522019169, + "act_iou_n_pairs": 6144, + "input_r2": 0.9657492789020655, + "matched_mcc_signed": { + "0.6": 0.831941015562173, + "0.7": 0.8528241095133843, + "0.8": 0.876032145109257, + "0.9": 0.9037045961017789 + }, + "matched_mcc_abs": { + "0.6": 0.8339820883279667, + "0.7": 0.8516776373880651, + "0.8": 0.8740829211955435, + "0.9": 0.9027055557863687 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 64.0, + "l0_std": 0.0, + "recon_r2_mean": 0.9198580042154813, + "recon_r2_std": 8.91968169242441e-05, + "l1_per_token_mean": 3.422150216634993, + "l1_per_token_std": 0.0027183972020539487, + "input_r2_mean": 0.9666731728020038, + "input_r2_std": 0.0008021041033293753, + "mcc_signed_mean": 0.438487440478341, + "mcc_signed_std": 0.0038484350062672613, + "mcc_abs_mean": 0.4446376325172279, + "mcc_abs_std": 0.003724907131902612, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8312444065353142, + "0.7": 0.8526350800921323, + "0.8": 0.8753306447338995, + "0.9": 0.9036850748142887 + }, + "matched_mcc_signed_band": 0.8657238015439087, + "matched_mcc_abs_per_threshold": { + "0.6": 0.835625040546041, + "0.7": 0.8532955136475383, + "0.8": 0.8753424870895535, + "0.9": 0.9031162460547915 + }, + "matched_mcc_abs_band": 0.8668448218344811, + "act_iou_mean": 0.258723443179113, + "act_iou_std": 0.0032192833929144704, + "act_iou_n_pairs": 6144.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/dict_metrics_step_final.json b/lm/pythia-160m_layer8/isae/x8/l0_64/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..912937e8137e2d26110c41e41ca882f54d77df16 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 6144, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.9202155766526814, + "l1_per_token": 3.42094151613428, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 79, + "dense_rate": 0.012858072916666666, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 4, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.15156280994415283, + "n_dup_abs": { + "0.5": 4, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 2, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 2, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.1611444056034088 + }, + "seed1": { + "d_sae": 6144, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.9203477760837526, + "l1_per_token": 3.415665264438531, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 79, + "dense_rate": 0.012858072916666666, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.15327702462673187, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.16284388303756714 + }, + "seed2": { + "d_sae": 6144, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.9201451809824376, + "l1_per_token": 3.4181913625278555, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 81, + "dense_rate": 0.01318359375, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_signed": 0.15186917781829834, + "n_dup_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "n_dup_clusters_abs": { + "0.5": 0, + "0.6": 0, + "0.7": 0, + "0.8": 0, + "0.9": 0 + }, + "largest_dup_abs": { + "0.5": 1, + "0.6": 1, + "0.7": 1, + "0.8": 1, + "0.9": 1 + }, + "babel_mu1_abs": 0.16109567880630493 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.49634912610054016, + "dcs_abs": 0.8050420880317688, + "frac_sign_flipped": 0.5061848958333334, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.49638280272483826, + "0.6": 0.49634918570518494, + "0.7": 0.49634918570518494, + "0.8": 0.49634918570518494, + "0.9": 0.49634918570518494 + }, + "unique_frac_signed": { + "0.5": 0.9998372395833334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.49630201125304063, + "0.6": 0.49634918570518494, + "0.7": 0.49634918570518494, + "0.8": 0.49634918570518494, + "0.9": 0.49634918570518494 + }, + "matched_frac_signed": { + "0.5": 0.4560546875, + "0.6": 0.42626953125, + "0.7": 0.3873697916666667, + "0.8": 0.337890625, + "0.9": 0.24820963541666666 + }, + "matched_cos_abs": { + "0.5": 0.8051762580871582, + "0.6": 0.8050422072410583, + "0.7": 0.8050422072410583, + "0.8": 0.8050422072410583, + "0.9": 0.8050422072410583 + }, + "unique_frac_abs": { + "0.5": 0.9998372395833334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.8050452072639018, + "0.6": 0.8050422072410583, + "0.7": 0.8050422072410583, + "0.8": 0.8050422072410583, + "0.9": 0.8050422072410583 + }, + "matched_frac_abs": { + "0.5": 0.8818359375, + "0.6": 0.8504231770833334, + "0.7": 0.7908528645833334, + "0.8": 0.6941731770833334, + "0.9": 0.51318359375 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.5024977922439575, + "dcs_abs": 0.802866518497467, + "frac_sign_flipped": 0.4986979166666667, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.5025244951248169, + "0.6": 0.5024977922439575, + "0.7": 0.5024977922439575, + "0.8": 0.5024977922439575, + "0.9": 0.5024977922439575 + }, + "unique_frac_signed": { + "0.5": 0.9998372395833334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.5024427040286052, + "0.6": 0.5024977922439575, + "0.7": 0.5024977922439575, + "0.8": 0.5024977922439575, + "0.9": 0.5024977922439575 + }, + "matched_frac_signed": { + "0.5": 0.4695638020833333, + "0.6": 0.4358723958333333, + "0.7": 0.3968098958333333, + "0.8": 0.3427734375, + "0.9": 0.2532552083333333 + }, + "matched_cos_abs": { + "0.5": 0.803025484085083, + "0.6": 0.802866518497467, + "0.7": 0.802866518497467, + "0.8": 0.802866518497467, + "0.9": 0.802866518497467 + }, + "unique_frac_abs": { + "0.5": 0.9998372395833334, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.8028947833226994, + "0.6": 0.802866518497467, + "0.7": 0.802866518497467, + "0.8": 0.802866518497467, + "0.9": 0.802866518497467 + }, + "matched_frac_abs": { + "0.5": 0.8811848958333334, + "0.6": 0.8448893229166666, + "0.7": 0.7874348958333334, + "0.8": 0.6827799479166666, + "0.9": 0.5126953125 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.5045285820960999, + "dcs_abs": 0.7998383045196533, + "frac_sign_flipped": 0.4982096354166667, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.5045286417007446, + "0.6": 0.5045286417007446, + "0.7": 0.5045286417007446, + "0.8": 0.5045286417007446, + "0.9": 0.5045286417007446 + }, + "unique_frac_signed": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.5045286417007446, + "0.6": 0.5045286417007446, + "0.7": 0.5045286417007446, + "0.8": 0.5045286417007446, + "0.9": 0.5045286417007446 + }, + "matched_frac_signed": { + "0.5": 0.4679361979166667, + "0.6": 0.43212890625, + "0.7": 0.3968098958333333, + "0.8": 0.3460286458333333, + "0.9": 0.2561848958333333 + }, + "matched_cos_abs": { + "0.5": 0.7998383045196533, + "0.6": 0.7998383045196533, + "0.7": 0.7998383045196533, + "0.8": 0.7998383045196533, + "0.9": 0.7998383045196533 + }, + "unique_frac_abs": { + "0.5": 1.0, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "dedup_dcs_abs": { + "0.5": 0.7998383045196533, + "0.6": 0.7998383045196533, + "0.7": 0.7998383045196533, + "0.8": 0.7998383045196533, + "0.9": 0.7998383045196533 + }, + "matched_frac_abs": { + "0.5": 0.8811848958333334, + "0.6": 0.8406575520833334, + "0.7": 0.78466796875, + "0.8": 0.6818033854166666, + "0.9": 0.5026041666666666 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 80, + "n_dense_std": 1.1547005383792517, + "l0_mean": 64, + "l0_std": 0.0, + "recon_r2_mean": 0.9202361779062905, + "recon_r2_std": null, + "l1_per_token_mean": 3.418266047700222, + "l1_per_token_std": 0.0026389200765050444, + "babel_mu1_signed_mean": 0.152236337463061, + "babel_mu1_signed_std": 0.0009141868780774971, + "n_dup_signed_per_threshold": { + "0.5": 1.3333333333333333, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_signed_band": 0.0, + "n_dup_signed_band_std": 0.0, + "n_dup_clusters_signed_per_threshold": { + "0.5": 0.6666666666666666, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_signed_band": 0.0, + "n_dup_clusters_signed_band_std": 0.0, + "largest_dup_signed_per_threshold": { + "0.5": 1.3333333333333333, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_signed_band": 1.0, + "largest_dup_signed_band_std": 0.0, + "dcs_signed_mean": 0.5011251668135325, + "dcs_signed_std": 0.004258984682410989, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.5010911189941302, + "0.6": 0.5011252065499624, + "0.7": 0.5011252065499624, + "0.8": 0.5011252065499624, + "0.9": 0.5011252065499624 + }, + "dedup_dcs_signed_band": 0.5011252065499624, + "dedup_dcs_signed_band_std": 0.004258975077567897, + "matched_cos_signed_per_threshold": { + "0.5": 0.5011453131834666, + "0.6": 0.5011252065499624, + "0.7": 0.5011252065499624, + "0.8": 0.5011252065499624, + "0.9": 0.5011252065499624 + }, + "matched_cos_signed_band": 0.5011252065499624, + "matched_cos_signed_band_std": 0.004258975077567897, + "unique_frac_signed_per_threshold": { + "0.5": 0.9998914930555557, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_signed_band": 1.0, + "unique_frac_signed_band_std": 0.0, + "matched_frac_signed_per_threshold": { + "0.5": 0.4645182291666667, + "0.6": 0.4314236111111111, + "0.7": 0.3936631944444444, + "0.8": 0.34223090277777773, + "0.9": 0.2525499131944444 + }, + "matched_frac_signed_band": 0.3549669053819444, + "matched_frac_signed_band_std": 0.004368520611196212, + "babel_mu1_abs_mean": 0.16169465581576029, + "babel_mu1_abs_std": 0.0009955581252219323, + "n_dup_abs_per_threshold": { + "0.5": 1.3333333333333333, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_abs_band": 0.0, + "n_dup_abs_band_std": 0.0, + "n_dup_clusters_abs_per_threshold": { + "0.5": 0.6666666666666666, + "0.6": 0.0, + "0.7": 0.0, + "0.8": 0.0, + "0.9": 0.0 + }, + "n_dup_clusters_abs_band": 0.0, + "n_dup_clusters_abs_band_std": 0.0, + "largest_dup_abs_per_threshold": { + "0.5": 1.3333333333333333, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "largest_dup_abs_band": 1.0, + "largest_dup_abs_band_std": 0.0, + "dcs_abs_mean": 0.802582303682963, + "dcs_abs_std": 0.002613508036304629, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.8025927650354182, + "0.6": 0.8025823434193929, + "0.7": 0.8025823434193929, + "0.8": 0.8025823434193929, + "0.9": 0.8025823434193929 + }, + "dedup_dcs_abs_band": 0.8025823434193929, + "dedup_dcs_abs_band_std": 0.0026135641353734747, + "matched_cos_abs_per_threshold": { + "0.5": 0.8026800155639648, + "0.6": 0.8025823434193929, + "0.7": 0.8025823434193929, + "0.8": 0.8025823434193929, + "0.9": 0.8025823434193929 + }, + "matched_cos_abs_band": 0.8025823434193929, + "matched_cos_abs_band_std": 0.0026135641353734747, + "unique_frac_abs_per_threshold": { + "0.5": 0.9998914930555557, + "0.6": 1.0, + "0.7": 1.0, + "0.8": 1.0, + "0.9": 1.0 + }, + "unique_frac_abs_band": 1.0, + "unique_frac_abs_band_std": 0.0, + "matched_frac_abs_per_threshold": { + "0.5": 0.8814019097222223, + "0.6": 0.8453233506944445, + "0.7": 0.7876519097222223, + "0.8": 0.6862521701388888, + "0.9": 0.5094943576388888 + }, + "matched_frac_abs_band": 0.7071804470486112, + "matched_frac_abs_band_std": 0.004866565943826758, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/seed0/config.json b/lm/pythia-160m_layer8/isae/x8/l0_64/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..912b65aaadacd5e52a6b294a895f07ff548f9a07 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/seed0/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 64, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..60651ff95cc4d9c0ffd1124ef816b35165b176b0 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 63.945493212890625, + "l0_per_token_0.05": 17.339835400390626, + "l0_per_token_0.1": 5.406124169921875, + "l0_per_token_0.5": 0.371471142578125, + "l0_per_token_1.0": 0.01589375, + "l1_per_token": 3.42094151613428, + "r2_score": 0.9202155766526814, + "l2_loss": 0.0531181543531738 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/seed1/config.json b/lm/pythia-160m_layer8/isae/x8/l0_64/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..884c307e2edf20d0b9bc4dfe9176dc9311b8e1eb --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/seed1/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 64, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..dbc955af4faabf5c5860e161efdfd9a0089b0e40 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 63.941989208984374, + "l0_per_token_0.05": 17.22340869140625, + "l0_per_token_0.1": 5.40553583984375, + "l0_per_token_0.5": 0.37226884765625, + "l0_per_token_1.0": 0.0156732421875, + "l1_per_token": 3.415665264438531, + "r2_score": 0.9203477760837526, + "l2_loss": 0.053030139807346055 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/seed2/config.json b/lm/pythia-160m_layer8/isae/x8/l0_64/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e0f0b8b967544c630769a9f6b12e5881dec0c6f2 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/seed2/config.json @@ -0,0 +1,20 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "isae", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 64, + "top_k_inference": 0, + "use_abstopk": true, + "num_encode_steps": 1, + "initial_step_size": 0.9, + "rip_weight": 0.1, + "rip_multiplier": 2, + "rip_weighted": false, + "l1_weight": 0.01, + "input_norm": true, + "class_type": "IsaeSAE", + "module_name": "consdl.models.isae" +} diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/isae/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..f33a4d109929091d6c007a86b4aa5fa8a81c0612 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 63.95524775390625, + "l0_per_token_0.05": 17.26467236328125, + "l0_per_token_0.1": 5.375287109375, + "l0_per_token_0.5": 0.369231298828125, + "l0_per_token_1.0": 0.01618115234375, + "l1_per_token": 3.4181913625278555, + "r2_score": 0.9201451809824376, + "l2_loss": 0.05316502174809786 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/isae/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/isae/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/isae/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/code_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x16/l0_128/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..25e64637ed5cf930f3c68e6fa9e78a8842a78ad4 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 130.37761, + "recon_r2": 0.9176537694930235, + "l1_per_token": 42.9301933673927 + }, + "seed1": { + "l0": 131.075455, + "recon_r2": 0.916804660461032, + "l1_per_token": 42.937659527870466 + }, + "seed2": { + "l0": 130.06846666666667, + "recon_r2": 0.9175106673914764, + "l1_per_token": 42.77485469381417 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.6741032671787422, + "mcc_abs": 0.6741061899289201, + "act_iou": 0.363188760993625, + "act_iou_n_pairs": 12288, + "input_r2": 0.9809421989882292, + "matched_mcc_signed": { + "0.6": 0.8091543836924385, + "0.7": 0.8607631631441156, + "0.8": 0.9059855704316839, + "0.9": 0.9452274187254209 + }, + "matched_mcc_abs": { + "0.6": 0.8086801756150358, + "0.7": 0.8603047626323993, + "0.8": 0.9058501023865029, + "0.9": 0.9453231624247234 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.6774748456811954, + "mcc_abs": 0.6774760763870386, + "act_iou": 0.3668046327647856, + "act_iou_n_pairs": 12288, + "input_r2": 0.9818270718944893, + "matched_mcc_signed": { + "0.6": 0.8144815665450736, + "0.7": 0.864134412218179, + "0.8": 0.9076120240710006, + "0.9": 0.9467939893935482 + }, + "matched_mcc_abs": { + "0.6": 0.8137459176945853, + "0.7": 0.8638708982407767, + "0.8": 0.9075548347536396, + "0.9": 0.9468001009581425 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.67725762293006, + "mcc_abs": 0.6772581414338655, + "act_iou": 0.3655596413892555, + "act_iou_n_pairs": 12288, + "input_r2": 0.9810731581940731, + "matched_mcc_signed": { + "0.6": 0.8102495913804436, + "0.7": 0.8648502858012179, + "0.8": 0.9071163592084736, + "0.9": 0.9462426905128144 + }, + "matched_mcc_abs": { + "0.6": 0.8096877737976645, + "0.7": 0.8641739760986913, + "0.8": 0.9069438765432971, + "0.9": 0.9462171301645778 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 130.50717722222223, + "l0_std": 0.5158460280676308, + "recon_r2_mean": 0.9173230324485107, + "recon_r2_std": 0.00045458958514100296, + "l1_per_token_mean": 42.88090252969244, + "l1_per_token_std": 0.09191595894576178, + "input_r2_mean": 0.9812808096922638, + "input_r2_std": 0.0004775870324704335, + "mcc_signed_mean": 0.676278578596666, + "mcc_signed_std": 0.0018870032471128432, + "mcc_abs_mean": 0.6762801359166081, + "mcc_abs_std": 0.00188584325408874, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8112951805393186, + "0.7": 0.8632492870545042, + "0.8": 0.9069046512370527, + "0.9": 0.9460880328772611 + }, + "matched_mcc_signed_band": 0.8818842879270341, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8107046223690952, + "0.7": 0.8627832123239557, + "0.8": 0.90678293789448, + "0.9": 0.9461134645158146 + }, + "matched_mcc_abs_band": 0.8815960592758364, + "act_iou_mean": 0.36518434504922204, + "act_iou_std": 0.0018369179822465844, + "act_iou_n_pairs": 12288.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/dict_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x16/l0_128/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..3fdc30cfbc7581db0f5c757ee265746076a3d893 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 12288, + "d_in": 768, + "l0": 130, + "l0_raw": 130.30217734375, + "recon_r2": 0.9180541109182091, + "l1_per_token": 42.82666226793812, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 29, + "dense_rate": 0.0023600260416666665, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 3614, + "0.6": 1407, + "0.7": 435, + "0.8": 92, + "0.9": 16 + }, + "n_dup_clusters_signed": { + "0.5": 1565, + "0.6": 623, + "0.7": 198, + "0.8": 40, + "0.9": 5 + }, + "largest_dup_signed": { + "0.5": 8, + "0.6": 8, + "0.7": 8, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_signed": 0.47603535652160645, + "n_dup_abs": { + "0.5": 3659, + "0.6": 1425, + "0.7": 438, + "0.8": 94, + "0.9": 16 + }, + "n_dup_clusters_abs": { + "0.5": 1583, + "0.6": 629, + "0.7": 200, + "0.8": 41, + "0.9": 5 + }, + "largest_dup_abs": { + "0.5": 12, + "0.6": 9, + "0.7": 7, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_abs": 0.48382917046546936 + }, + "seed1": { + "d_sae": 12288, + "d_in": 768, + "l0": 131, + "l0_raw": 130.9643427734375, + "recon_r2": 0.9172755785292012, + "l1_per_token": 42.81743071048249, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 28, + "dense_rate": 0.0022786458333333335, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 3703, + "0.6": 1426, + "0.7": 438, + "0.8": 90, + "0.9": 12 + }, + "n_dup_clusters_signed": { + "0.5": 1595, + "0.6": 633, + "0.7": 197, + "0.8": 41, + "0.9": 4 + }, + "largest_dup_signed": { + "0.5": 11, + "0.6": 7, + "0.7": 7, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_signed": 0.47662830352783203, + "n_dup_abs": { + "0.5": 3748, + "0.6": 1437, + "0.7": 438, + "0.8": 92, + "0.9": 12 + }, + "n_dup_clusters_abs": { + "0.5": 1615, + "0.6": 636, + "0.7": 195, + "0.8": 42, + "0.9": 4 + }, + "largest_dup_abs": { + "0.5": 15, + "0.6": 7, + "0.7": 7, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_abs": 0.48444533348083496 + }, + "seed2": { + "d_sae": 12288, + "d_in": 768, + "l0": 130, + "l0_raw": 130.06357700195312, + "recon_r2": 0.917967018501029, + "l1_per_token": 42.68159866125871, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 33, + "dense_rate": 0.002685546875, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 3642, + "0.6": 1441, + "0.7": 425, + "0.8": 90, + "0.9": 10 + }, + "n_dup_clusters_signed": { + "0.5": 1573, + "0.6": 645, + "0.7": 194, + "0.8": 41, + "0.9": 3 + }, + "largest_dup_signed": { + "0.5": 9, + "0.6": 7, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_signed": 0.47644317150115967, + "n_dup_abs": { + "0.5": 3677, + "0.6": 1455, + "0.7": 429, + "0.8": 94, + "0.9": 10 + }, + "n_dup_clusters_abs": { + "0.5": 1590, + "0.6": 650, + "0.7": 195, + "0.8": 43, + "0.9": 3 + }, + "largest_dup_abs": { + "0.5": 10, + "0.6": 8, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_abs": 0.48415717482566833 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.7970147728919983, + "dcs_abs": 0.7972627282142639, + "frac_sign_flipped": 0.008382161458333334, + "alive_source": "stats", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.5": 0.7801938056945801, + "0.6": 0.788032591342926, + "0.7": 0.7939623594284058, + "0.8": 0.7962612509727478, + "0.9": 0.797027587890625 + }, + "unique_frac_signed": { + "0.5": 0.8308512369791666, + "0.6": 0.9358317057291666, + "0.7": 0.9805501302083334, + "0.8": 0.9958902994791666, + "0.9": 0.9992268880208334 + }, + "dedup_dcs_signed": { + "0.5": 0.6482249885448255, + "0.6": 0.7374658841266257, + "0.7": 0.7785198949180389, + "0.8": 0.7929888556949056, + "0.9": 0.7964113963147005 + }, + "matched_frac_signed": { + "0.5": 0.6834309895833334, + "0.6": 0.7063802083333334, + "0.7": 0.6796875, + "0.8": 0.6117350260416666, + "0.9": 0.5040690104166666 + }, + "matched_cos_abs": { + "0.5": 0.7804387211799622, + "0.6": 0.7880868911743164, + "0.7": 0.7941667437553406, + "0.8": 0.7964543104171753, + "0.9": 0.7972420454025269 + }, + "unique_frac_abs": { + "0.5": 0.8287353515625, + "0.6": 0.9350179036458334, + "0.7": 0.9804280598958334, + "0.8": 0.9958089192708334, + "0.9": 0.9992268880208334 + }, + "dedup_dcs_abs": { + "0.5": 0.6467771579700639, + "0.6": 0.7368753528765714, + "0.7": 0.77862335981384, + "0.8": 0.7931163061051242, + "0.9": 0.7966256880269309 + }, + "matched_frac_abs": { + "0.5": 0.681640625, + "0.6": 0.7059733072916666, + "0.7": 0.6800130208333334, + "0.8": 0.6111653645833334, + "0.9": 0.5038248697916666 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.7985073924064636, + "dcs_abs": 0.7987324595451355, + "frac_sign_flipped": 0.00732421875, + "alive_source": "stats", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.5": 0.7809359431266785, + "0.6": 0.7905004024505615, + "0.7": 0.7952756881713867, + "0.8": 0.7979108691215515, + "0.9": 0.7985433340072632 + }, + "unique_frac_signed": { + "0.5": 0.8324381510416666, + "0.6": 0.9357096354166666, + "0.7": 0.98095703125, + "0.8": 0.9958902994791666, + "0.9": 0.999267578125 + }, + "dedup_dcs_signed": { + "0.5": 0.6500808725783523, + "0.6": 0.7396788433737431, + "0.7": 0.7801312780939043, + "0.8": 0.7946316944071441, + "0.9": 0.7979584634013008 + }, + "matched_frac_signed": { + "0.5": 0.682373046875, + "0.6": 0.7063802083333334, + "0.7": 0.6805013020833334, + "0.8": 0.6170247395833334, + "0.9": 0.50927734375 + }, + "matched_cos_abs": { + "0.5": 0.7810721397399902, + "0.6": 0.7905446290969849, + "0.7": 0.7954003214836121, + "0.8": 0.798076331615448, + "0.9": 0.7987546324729919 + }, + "unique_frac_abs": { + "0.5": 0.8306070963541666, + "0.6": 0.9348551432291666, + "0.7": 0.9807942708333334, + "0.8": 0.9957682291666666, + "0.9": 0.999267578125 + }, + "dedup_dcs_abs": { + "0.5": 0.6487640620325692, + "0.6": 0.7390447124635102, + "0.7": 0.7801240783301182, + "0.8": 0.794699055472544, + "0.9": 0.7981696071074111 + }, + "matched_frac_abs": { + "0.5": 0.6803385416666666, + "0.6": 0.7061360677083334, + "0.7": 0.6807454427083334, + "0.8": 0.6171875, + "0.9": 0.5091959635416666 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.7988874316215515, + "dcs_abs": 0.7991248965263367, + "frac_sign_flipped": 0.006998697916666667, + "alive_source": "stats", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.5": 0.7797626256942749, + "0.6": 0.7892056107521057, + "0.7": 0.7957267761230469, + "0.8": 0.7981976866722107, + "0.9": 0.7988100051879883 + }, + "unique_frac_signed": { + "0.5": 0.8300374348958334, + "0.6": 0.9353434244791666, + "0.7": 0.9807942708333334, + "0.8": 0.9960123697916666, + "0.9": 0.9993896484375 + }, + "dedup_dcs_signed": { + "0.5": 0.6472321696589158, + "0.6": 0.7381782785790467, + "0.7": 0.7804442631701629, + "0.8": 0.7950147694646148, + "0.9": 0.7983224502531812 + }, + "matched_frac_signed": { + "0.5": 0.680908203125, + "0.6": 0.7078450520833334, + "0.7": 0.67724609375, + "0.8": 0.6142578125, + "0.9": 0.5056966145833334 + }, + "matched_cos_abs": { + "0.5": 0.7804279923439026, + "0.6": 0.7893691062927246, + "0.7": 0.7959301471710205, + "0.8": 0.7984463572502136, + "0.9": 0.7990484237670898 + }, + "unique_frac_abs": { + "0.5": 0.8282877604166666, + "0.6": 0.9346516927083334, + "0.7": 0.9805908203125, + "0.8": 0.9958902994791666, + "0.9": 0.9993896484375 + }, + "dedup_dcs_abs": { + "0.5": 0.6464189539450065, + "0.6": 0.7377851713681594, + "0.7": 0.7804817959258799, + "0.8": 0.7951649818399649, + "0.9": 0.7985607233131304 + }, + "matched_frac_abs": { + "0.5": 0.6796061197916666, + "0.6": 0.7074381510416666, + "0.7": 0.6778971354166666, + "0.8": 0.6142578125, + "0.9": 0.5057779947916666 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 30, + "n_dense_std": 2.6457513110645907, + "l0_mean": 130, + "l0_std": 0.4666855033918535, + "recon_r2_mean": 0.9177655693161464, + "recon_r2_std": null, + "l1_per_token_mean": 42.775230546559776, + "l1_per_token_std": 0.08121885785972129, + "babel_mu1_signed_mean": 0.4763689438501994, + "babel_mu1_signed_std": 0.00030336256554374476, + "n_dup_signed_per_threshold": { + "0.5": 3653.0, + "0.6": 1424.6666666666667, + "0.7": 432.6666666666667, + "0.8": 90.66666666666667, + "0.9": 12.666666666666666 + }, + "n_dup_signed_band": 490.16666666666674, + "n_dup_signed_band_std": 2.309401076758503, + "n_dup_clusters_signed_per_threshold": { + "0.5": 1577.6666666666667, + "0.6": 633.6666666666666, + "0.7": 196.33333333333334, + "0.8": 40.666666666666664, + "0.9": 4.0 + }, + "n_dup_clusters_signed_band": 218.66666666666666, + "n_dup_clusters_signed_band_std": 2.126225137028846, + "largest_dup_signed_per_threshold": { + "0.5": 9.333333333333334, + "0.6": 7.333333333333333, + "0.7": 7.0, + "0.8": 5.666666666666667, + "0.9": 5.666666666666667 + }, + "largest_dup_signed_band": 6.416666666666667, + "largest_dup_signed_band_std": 0.9464847243000456, + "dcs_signed_mean": 0.7981365323066711, + "dcs_signed_std": 0.0009898816038075155, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.6485126769273645, + "0.6": 0.7384410020264719, + "0.7": 0.7796984787273686, + "0.8": 0.7942117731888881, + "0.9": 0.7975641033230608 + }, + "dedup_dcs_signed_band": 0.7774788393164473, + "dedup_dcs_signed_band_std": 0.0009821726850709855, + "matched_cos_signed_per_threshold": { + "0.5": 0.7802974581718445, + "0.6": 0.7892462015151978, + "0.7": 0.7949882745742798, + "0.8": 0.7974566022555033, + "0.9": 0.7981269756952921 + }, + "matched_cos_signed_band": 0.7949545135100682, + "matched_cos_signed_band_std": 0.0009823670859045084, + "unique_frac_signed_per_threshold": { + "0.5": 0.8311089409722222, + "0.6": 0.9356282552083334, + "0.7": 0.9807671440972223, + "0.8": 0.9959309895833334, + "0.9": 0.9992947048611112 + }, + "unique_frac_signed_band": 0.9779052734375001, + "unique_frac_signed_band_std": 4.434101301612686e-05, + "matched_frac_signed_per_threshold": { + "0.5": 0.6822374131944445, + "0.6": 0.7068684895833334, + "0.7": 0.6791449652777778, + "0.8": 0.6143391927083334, + "0.9": 0.50634765625 + }, + "matched_frac_signed_band": 0.6266750759548612, + "matched_frac_signed_band_std": 0.0014586612544661623, + "babel_mu1_abs_mean": 0.4841438929239909, + "babel_mu1_abs_std": 0.00030829615965858936, + "n_dup_abs_per_threshold": { + "0.5": 3694.6666666666665, + "0.6": 1439.0, + "0.7": 435.0, + "0.8": 93.33333333333333, + "0.9": 12.666666666666666 + }, + "n_dup_abs_band": 495.0, + "n_dup_abs_band_std": 1.8874586088176875, + "n_dup_clusters_abs_per_threshold": { + "0.5": 1596.0, + "0.6": 638.3333333333334, + "0.7": 196.66666666666666, + "0.8": 42.0, + "0.9": 4.0 + }, + "n_dup_clusters_abs_band": 220.25, + "n_dup_clusters_abs_band_std": 2.179449471770337, + "largest_dup_abs_per_threshold": { + "0.5": 12.333333333333334, + "0.6": 8.0, + "0.7": 6.666666666666667, + "0.8": 5.666666666666667, + "0.9": 5.666666666666667 + }, + "largest_dup_abs_band": 6.500000000000001, + "largest_dup_abs_band_std": 0.8660254037844386, + "dcs_abs_mean": 0.7983733614285787, + "dcs_abs_std": 0.0009816472373556373, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.6473200579825465, + "0.6": 0.7379017455694137, + "0.7": 0.7797430780232794, + "0.8": 0.794326781139211, + "0.9": 0.7977853394824909 + }, + "dedup_dcs_abs_band": 0.7774392360535987, + "dedup_dcs_abs_band_std": 0.0009778101000459977, + "matched_cos_abs_per_threshold": { + "0.5": 0.780646284421285, + "0.6": 0.7893335421880087, + "0.7": 0.7951657374699911, + "0.8": 0.7976589997609457, + "0.9": 0.7983483672142029 + }, + "matched_cos_abs_band": 0.795126661658287, + "matched_cos_abs_band_std": 0.0009865475379558803, + "unique_frac_abs_per_threshold": { + "0.5": 0.8292100694444443, + "0.6": 0.9348415798611112, + "0.7": 0.9806043836805557, + "0.8": 0.9958224826388888, + "0.9": 0.9992947048611112 + }, + "unique_frac_abs_band": 0.9776407877604167, + "unique_frac_abs_band_std": 2.6913974111543305e-05, + "matched_frac_abs_per_threshold": { + "0.5": 0.6805284288194443, + "0.6": 0.7065158420138888, + "0.7": 0.6795518663194445, + "0.8": 0.6142035590277778, + "0.9": 0.5062662760416666 + }, + "matched_frac_abs_band": 0.6266343858506944, + "matched_frac_abs_band_std": 0.001556673489300874, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/config.json b/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5949eb30b783de244a80f491fa25f7af3f2edec1 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/lock.lock b/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..685775cd143c1e9112467f726482c19988153b0b --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 130.30217734375, + "l0_per_token_0.001": 129.70591313476564, + "l0_per_token_0.01": 124.46754379882813, + "l0_per_token_0.05": 103.85901723632813, + "l0_per_token_0.1": 83.34994228515625, + "l0_per_token_0.5": 20.833939404296874, + "l0_per_token_1.0": 7.666075244140625, + "l1_per_token": 42.82666226793812, + "r2_score": 0.9180541109182091, + "l2_loss": 0.05455719553058498 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/config.json b/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4c15515c994745470445ff31f3f2293062c96289 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/lock.lock b/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..451f4ed4c28eee9ca309d3fb7af88d47d0bd456d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 130.9643427734375, + "l0_per_token_0.001": 130.3665798828125, + "l0_per_token_0.01": 125.11216000976563, + "l0_per_token_0.05": 104.43537392578125, + "l0_per_token_0.1": 83.84158559570312, + "l0_per_token_0.5": 20.921650634765626, + "l0_per_token_1.0": 7.646245556640625, + "l1_per_token": 42.81743071048249, + "r2_score": 0.9172755785292012, + "l2_loss": 0.055075519808348364 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/config.json b/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..43913a5027e3e99b474c9dfef9490cb3c811038a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/lock.lock b/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..40be8b6cc237fb115397e49b99bce5ea31e1b5b1 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 130.06357700195312, + "l0_per_token_0.001": 129.46941254882813, + "l0_per_token_0.01": 124.25140581054687, + "l0_per_token_0.05": 103.71781689453125, + "l0_per_token_0.1": 83.27899145507813, + "l0_per_token_0.5": 20.845399365234375, + "l0_per_token_1.0": 7.617060888671875, + "l1_per_token": 42.68159866125871, + "r2_score": 0.917967018501029, + "l2_loss": 0.054615179135211925 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/code_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x16/l0_32/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..726360a58c45523f3b64f11a256ae22bd2fd757f --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 32.82785666666667, + "recon_r2": 0.8628697903856377, + "l1_per_token": 25.778066864200742 + }, + "seed1": { + "l0": 32.90056333333333, + "recon_r2": 0.8624777672669597, + "l1_per_token": 25.724271299599273 + }, + "seed2": { + "l0": 32.344076666666666, + "recon_r2": 0.862130650117375, + "l1_per_token": 25.54572037447116 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.6849664710366296, + "mcc_abs": 0.6849694113016055, + "act_iou": 0.40898386200867476, + "act_iou_n_pairs": 11902, + "input_r2": 0.9813973823586749, + "matched_mcc_signed": { + "0.6": 0.8034289487627333, + "0.7": 0.8533715664160366, + "0.8": 0.8960160892489137, + "0.9": 0.9378890461879259 + }, + "matched_mcc_abs": { + "0.6": 0.802930457713376, + "0.7": 0.8531028182491096, + "0.8": 0.8961074415607467, + "0.9": 0.937849249914135 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.684873600773641, + "mcc_abs": 0.6848757268187255, + "act_iou": 0.41116722827974944, + "act_iou_n_pairs": 11880, + "input_r2": 0.9817382881802152, + "matched_mcc_signed": { + "0.6": 0.8041145819697469, + "0.7": 0.8541622982325018, + "0.8": 0.8961054821891907, + "0.9": 0.9381123800931228 + }, + "matched_mcc_abs": { + "0.6": 0.8044189848649197, + "0.7": 0.853505129694458, + "0.8": 0.8963777377131, + "0.9": 0.9380919267508103 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.688733309956085, + "mcc_abs": 0.6887365086772101, + "act_iou": 0.415440367469769, + "act_iou_n_pairs": 11820, + "input_r2": 0.981645957806895, + "matched_mcc_signed": { + "0.6": 0.8069601112244396, + "0.7": 0.8551893831409293, + "0.8": 0.8944591046046382, + "0.9": 0.9380651335120552 + }, + "matched_mcc_abs": { + "0.6": 0.8060998539136016, + "0.7": 0.8551149910839805, + "0.8": 0.8945603280621693, + "0.9": 0.9380155493401472 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 32.69083222222222, + "l0_std": 0.30249153099868087, + "recon_r2_mean": 0.8624927359233241, + "recon_r2_std": 0.000369797416901319, + "l1_per_token_mean": 25.682686179423726, + "l1_per_token_std": 0.12162735912378386, + "input_r2_mean": 0.9815938761152617, + "input_r2_std": 0.00017631951605696374, + "mcc_signed_mean": 0.6861911272554518, + "mcc_signed_std": 0.00220208444070285, + "mcc_abs_mean": 0.686193882265847, + "mcc_abs_std": 0.002202477240836771, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8048345473189733, + "0.7": 0.8542410825964892, + "0.8": 0.8955268920142476, + "0.9": 0.9380221865977013 + }, + "matched_mcc_signed_band": 0.8731561771318529, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8044830988306324, + "0.7": 0.853907646342516, + "0.8": 0.8956818357786721, + "0.9": 0.9379855753350309 + }, + "matched_mcc_abs_band": 0.8730145390717129, + "act_iou_mean": 0.411863819252731, + "act_iou_std": 0.003284135339783438, + "act_iou_n_pairs": 11867.333333333334, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/dict_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x16/l0_32/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..0b4f3f9c9a9547da6cc4088ebb2470cbb3641070 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 12288, + "d_in": 768, + "l0": 33, + "l0_raw": 32.83185390625, + "recon_r2": 0.8635755356361929, + "l1_per_token": 25.715061958785956, + "stats_source": "stats", + "n_dead": 745, + "dead_rate": 0.060628255208333336, + "alive_source": "stats", + "n_dense": 8, + "dense_rate": 0.0006510416666666666, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 7245, + "0.6": 3799, + "0.7": 1322, + "0.8": 264, + "0.9": 14 + }, + "n_dup_clusters_signed": { + "0.5": 2949, + "0.6": 1670, + "0.7": 595, + "0.8": 123, + "0.9": 5 + }, + "largest_dup_signed": { + "0.5": 14, + "0.6": 8, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_signed": 0.5475583076477051, + "n_dup_abs": { + "0.5": 7263, + "0.6": 3816, + "0.7": 1332, + "0.8": 264, + "0.9": 14 + }, + "n_dup_clusters_abs": { + "0.5": 2955, + "0.6": 1677, + "0.7": 599, + "0.8": 123, + "0.9": 5 + }, + "largest_dup_abs": { + "0.5": 14, + "0.6": 8, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_abs": 0.5484992265701294 + }, + "seed1": { + "d_sae": 12288, + "d_in": 768, + "l0": 33, + "l0_raw": 32.92201201171875, + "recon_r2": 0.8632641941767533, + "l1_per_token": 25.680720818849334, + "stats_source": "stats", + "n_dead": 679, + "dead_rate": 0.055257161458333336, + "alive_source": "stats", + "n_dense": 10, + "dense_rate": 0.0008138020833333334, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 7265, + "0.6": 3809, + "0.7": 1342, + "0.8": 301, + "0.9": 20 + }, + "n_dup_clusters_signed": { + "0.5": 2952, + "0.6": 1666, + "0.7": 608, + "0.8": 140, + "0.9": 8 + }, + "largest_dup_signed": { + "0.5": 10, + "0.6": 8, + "0.7": 5, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_signed": 0.5508561730384827, + "n_dup_abs": { + "0.5": 7273, + "0.6": 3838, + "0.7": 1352, + "0.8": 303, + "0.9": 20 + }, + "n_dup_clusters_abs": { + "0.5": 2955, + "0.6": 1680, + "0.7": 613, + "0.8": 141, + "0.9": 8 + }, + "largest_dup_abs": { + "0.5": 11, + "0.6": 8, + "0.7": 5, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_abs": 0.5517326593399048 + }, + "seed2": { + "d_sae": 12288, + "d_in": 768, + "l0": 32, + "l0_raw": 32.3768498046875, + "recon_r2": 0.8628500154977831, + "l1_per_token": 25.501685437012753, + "stats_source": "stats", + "n_dead": 702, + "dead_rate": 0.05712890625, + "alive_source": "stats", + "n_dense": 7, + "dense_rate": 0.0005696614583333334, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 7335, + "0.6": 3911, + "0.7": 1325, + "0.8": 309, + "0.9": 17 + }, + "n_dup_clusters_signed": { + "0.5": 2992, + "0.6": 1709, + "0.7": 596, + "0.8": 141, + "0.9": 5 + }, + "largest_dup_signed": { + "0.5": 13, + "0.6": 9, + "0.7": 9, + "0.8": 9, + "0.9": 9 + }, + "babel_mu1_signed": 0.552829384803772, + "n_dup_abs": { + "0.5": 7343, + "0.6": 3926, + "0.7": 1331, + "0.8": 307, + "0.9": 17 + }, + "n_dup_clusters_abs": { + "0.5": 2994, + "0.6": 1715, + "0.7": 599, + "0.8": 140, + "0.9": 5 + }, + "largest_dup_abs": { + "0.5": 13, + "0.6": 9, + "0.7": 9, + "0.8": 9, + "0.9": 9 + }, + "babel_mu1_abs": 0.55360347032547 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.8013023734092712, + "dcs_abs": 0.8019312024116516, + "frac_sign_flipped": 0.042887369791666664, + "alive_source": "stats", + "n_alive_a": 11543, + "n_alive_b": 11609, + "matched_cos_signed": { + "0.5": 0.8262660503387451, + "0.6": 0.8323167562484741, + "0.7": 0.8392307162284851, + "0.8": 0.8430661559104919, + "0.9": 0.8450356721878052 + }, + "unique_frac_signed": { + "0.5": 0.6281962681409814, + "0.6": 0.8155234968901175, + "0.7": 0.9368953006219765, + "0.8": 0.9869557705597789, + "0.9": 0.9990929509329648 + }, + "dedup_dcs_signed": { + "0.5": 0.519057249314388, + "0.6": 0.6787738715759951, + "0.7": 0.7862713141720832, + "0.8": 0.8320690075395103, + "0.9": 0.8442691833697358 + }, + "matched_frac_signed": { + "0.5": 0.539651002073255, + "0.6": 0.6828783690393918, + "0.7": 0.7485314443676572, + "0.8": 0.7290946786454734, + "0.9": 0.6356254319281271 + }, + "matched_cos_abs": { + "0.5": 0.8270618319511414, + "0.6": 0.8324286341667175, + "0.7": 0.8395044207572937, + "0.8": 0.8432877659797668, + "0.9": 0.8452879190444946 + }, + "unique_frac_abs": { + "0.5": 0.62746199032481, + "0.6": 0.8144436765722184, + "0.7": 0.9364201796821009, + "0.8": 0.9869125777470629, + "0.9": 0.9990929509329648 + }, + "dedup_dcs_abs": { + "0.5": 0.5189498631977467, + "0.6": 0.6779662372947316, + "0.7": 0.786128880529463, + "0.8": 0.8322513029056536, + "0.9": 0.8445212014261492 + }, + "matched_frac_abs": { + "0.5": 0.5399101589495507, + "0.6": 0.6824464409122322, + "0.7": 0.7479267449896337, + "0.8": 0.7294402211472011, + "0.9": 0.6360573600552868 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.8016643524169922, + "dcs_abs": 0.8023223876953125, + "frac_sign_flipped": 0.043538411458333336, + "alive_source": "stats", + "n_alive_a": 11543, + "n_alive_b": 11586, + "matched_cos_signed": { + "0.5": 0.8244373202323914, + "0.6": 0.8331458568572998, + "0.7": 0.839341938495636, + "0.8": 0.8428651690483093, + "0.9": 0.8454399704933167 + }, + "unique_frac_signed": { + "0.5": 0.6264862294089671, + "0.6": 0.8127459034113018, + "0.7": 0.9370487267067318, + "0.8": 0.9866401487310303, + "0.9": 0.9990920489428855 + }, + "dedup_dcs_signed": { + "0.5": 0.516498628136424, + "0.6": 0.6771358821048693, + "0.7": 0.7865042947388957, + "0.8": 0.831604615750029, + "0.9": 0.8446723523783803 + }, + "matched_frac_signed": { + "0.5": 0.5398417571014743, + "0.6": 0.6824333088330667, + "0.7": 0.7489299148255437, + "0.8": 0.7287820485105279, + "0.9": 0.6372951705650914 + }, + "matched_cos_abs": { + "0.5": 0.8246745467185974, + "0.6": 0.833436131477356, + "0.7": 0.8397386074066162, + "0.8": 0.8431859016418457, + "0.9": 0.8457317352294922 + }, + "unique_frac_abs": { + "0.5": 0.6257079856457262, + "0.6": 0.8119244238834363, + "0.7": 0.9366596048251113, + "0.8": 0.9866833844956548, + "0.9": 0.9990920489428855 + }, + "dedup_dcs_abs": { + "0.5": 0.5160054494405959, + "0.6": 0.6766871508933922, + "0.7": 0.7865492321698705, + "0.8": 0.8319575191909967, + "0.9": 0.8449638522064553 + }, + "matched_frac_abs": { + "0.5": 0.5394958709844784, + "0.6": 0.6816550650698258, + "0.7": 0.7490163863547927, + "0.8": 0.7284361623935319, + "0.9": 0.6373816420943405 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.8057601451873779, + "dcs_abs": 0.8063613772392273, + "frac_sign_flipped": 0.038492838541666664, + "alive_source": "stats", + "n_alive_a": 11609, + "n_alive_b": 11586, + "matched_cos_signed": { + "0.5": 0.832414984703064, + "0.6": 0.8367299437522888, + "0.7": 0.8415942192077637, + "0.8": 0.8458774089813232, + "0.9": 0.8470730185508728 + }, + "unique_frac_signed": { + "0.5": 0.6268592369045053, + "0.6": 0.8127182582453115, + "0.7": 0.936926061651218, + "0.8": 0.9858159086009916, + "0.9": 0.9989652942444492 + }, + "dedup_dcs_signed": { + "0.5": 0.5218070220988381, + "0.6": 0.6800257025080576, + "0.7": 0.7885115573107618, + "0.8": 0.8338794064999757, + "0.9": 0.8461965472232065 + }, + "matched_frac_signed": { + "0.5": 0.5415822375511964, + "0.6": 0.6804914852338866, + "0.7": 0.7511101530502263, + "0.8": 0.7335201552058633, + "0.9": 0.6372925199396422 + }, + "matched_cos_abs": { + "0.5": 0.8327586054801941, + "0.6": 0.8367635607719421, + "0.7": 0.8418205380439758, + "0.8": 0.8460549712181091, + "0.9": 0.8473203778266907 + }, + "unique_frac_abs": { + "0.5": 0.6263849967665445, + "0.6": 0.8116835524897608, + "0.7": 0.9365811597327011, + "0.8": 0.9858159086009916, + "0.9": 0.9989652942444492 + }, + "dedup_dcs_abs": { + "0.5": 0.5216274964010235, + "0.6": 0.6791872196013519, + "0.7": 0.7884332558080333, + "0.8": 0.834054450177766, + "0.9": 0.846443650554958 + }, + "matched_frac_abs": { + "0.5": 0.541064884673421, + "0.6": 0.6799741323561113, + "0.7": 0.7509377020909679, + "0.8": 0.733778831644751, + "0.9": 0.6375511963785299 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 709, + "n_dead_std": 33.50124375800596, + "n_dense_mean": 8, + "n_dense_std": 1.5275252316519468, + "l0_mean": 33, + "l0_std": 0.29222110589859496, + "recon_r2_mean": 0.8632299151035764, + "recon_r2_std": null, + "l1_per_token_mean": 25.63248940488268, + "l1_per_token_std": 0.11457350034333634, + "babel_mu1_signed_mean": 0.5504146218299866, + "babel_mu1_signed_std": 0.002663135219735535, + "n_dup_signed_per_threshold": { + "0.5": 7281.666666666667, + "0.6": 3839.6666666666665, + "0.7": 1329.6666666666667, + "0.8": 291.3333333333333, + "0.9": 17.0 + }, + "n_dup_signed_band": 1369.4166666666665, + "n_dup_signed_band_std": 20.411904206451034, + "n_dup_clusters_signed_per_threshold": { + "0.5": 2964.3333333333335, + "0.6": 1681.6666666666667, + "0.7": 599.6666666666666, + "0.8": 134.66666666666666, + "0.9": 6.0 + }, + "n_dup_clusters_signed_band": 605.5, + "n_dup_clusters_signed_band_std": 7.25, + "largest_dup_signed_per_threshold": { + "0.5": 12.333333333333334, + "0.6": 8.333333333333334, + "0.7": 6.666666666666667, + "0.8": 6.333333333333333, + "0.9": 6.333333333333333 + }, + "largest_dup_signed_band": 6.916666666666666, + "largest_dup_signed_band_std": 1.8085445345175588, + "dcs_signed_mean": 0.8029089570045471, + "dcs_signed_std": 0.0024758256685282194, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.51912096651655, + "0.6": 0.678645152062974, + "0.7": 0.7870957220739135, + "0.8": 0.8325176765965049, + "0.9": 0.8450460276571076 + }, + "dedup_dcs_signed_band": 0.785826144597625, + "dedup_dcs_signed_band_std": 0.0011638745695265534, + "matched_cos_signed_per_threshold": { + "0.5": 0.8277061184247335, + "0.6": 0.8340641856193542, + "0.7": 0.8400556246439616, + "0.8": 0.8439362446467081, + "0.9": 0.8458495537439982 + }, + "matched_cos_signed_band": 0.8409764021635054, + "matched_cos_signed_band_std": 0.0016018230797897768, + "unique_frac_signed_per_threshold": { + "0.5": 0.6271805781514846, + "0.6": 0.8136625528489102, + "0.7": 0.936956696326642, + "0.8": 0.9864706092972669, + "0.9": 0.9990500980400999 + }, + "unique_frac_signed_band": 0.9340349891282298, + "unique_frac_signed_band_std": 0.0005223970334319993, + "matched_frac_signed_per_threshold": { + "0.5": 0.5403583322419753, + "0.6": 0.681934387702115, + "0.7": 0.7495238374144758, + "0.8": 0.7304656274539548, + "0.9": 0.6367377074776203 + }, + "matched_frac_signed_band": 0.6996653900120414, + "matched_frac_signed_band_std": 0.0008288445763294788, + "babel_mu1_abs_mean": 0.5512784520785013, + "babel_mu1_abs_std": 0.002582257588955254, + "n_dup_abs_per_threshold": { + "0.5": 7293.0, + "0.6": 3860.0, + "0.7": 1338.3333333333333, + "0.8": 291.3333333333333, + "0.9": 17.0 + }, + "n_dup_abs_band": 1376.6666666666665, + "n_dup_abs_band_std": 19.423460899987248, + "n_dup_clusters_abs_per_threshold": { + "0.5": 2968.0, + "0.6": 1690.6666666666667, + "0.7": 603.6666666666666, + "0.8": 134.66666666666666, + "0.9": 6.0 + }, + "n_dup_clusters_abs_band": 608.75, + "n_dup_clusters_abs_band_std": 7.0400639201643616, + "largest_dup_abs_per_threshold": { + "0.5": 12.666666666666666, + "0.6": 8.333333333333334, + "0.7": 6.666666666666667, + "0.8": 6.333333333333333, + "0.9": 6.333333333333333 + }, + "largest_dup_abs_band": 6.916666666666666, + "largest_dup_abs_band_std": 1.8085445345175588, + "dcs_abs_mean": 0.8035383224487305, + "dcs_abs_std": 0.0024526486181572203, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.5188609363464554, + "0.6": 0.6779468692631586, + "0.7": 0.7870371228357889, + "0.8": 0.832754424091472, + "0.9": 0.8453095680625209 + }, + "dedup_dcs_abs_band": 0.7857619960632352, + "dedup_dcs_abs_band_std": 0.001101395552698971, + "matched_cos_abs_per_threshold": { + "0.5": 0.8281649947166443, + "0.6": 0.8342094421386719, + "0.7": 0.8403545220692953, + "0.8": 0.8441762129465739, + "0.9": 0.8461133440335592 + }, + "matched_cos_abs_band": 0.8412133802970251, + "matched_cos_abs_band_std": 0.0015511612786835772, + "unique_frac_abs_per_threshold": { + "0.5": 0.6265183242456935, + "0.6": 0.8126838843151386, + "0.7": 0.9365536480799711, + "0.8": 0.9864706236145698, + "0.9": 0.9990500980400999 + }, + "unique_frac_abs_band": 0.9336895635124449, + "unique_frac_abs_band_std": 0.0004856700713409624, + "matched_frac_abs_per_threshold": { + "0.5": 0.5401569715358167, + "0.6": 0.6813585461127231, + "0.7": 0.7492936111451315, + "0.8": 0.7305517383951613, + "0.9": 0.6369967328427191 + }, + "matched_frac_abs_band": 0.6995501571239338, + "matched_frac_abs_band_std": 0.0008783617965314018, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/config.json b/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5949eb30b783de244a80f491fa25f7af3f2edec1 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/lock.lock b/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..0c467188e41b2dc188bad093b257fca17d36c2f7 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.83185390625, + "l0_per_token_0.001": 32.7661697265625, + "l0_per_token_0.01": 32.18163466796875, + "l0_per_token_0.05": 29.739264501953127, + "l0_per_token_0.1": 27.007831884765626, + "l0_per_token_0.5": 13.636692919921876, + "l0_per_token_1.0": 6.9827642578125, + "l1_per_token": 25.715061958785956, + "r2_score": 0.8635755356361929, + "l2_loss": 0.09082745041697798 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/config.json b/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4c15515c994745470445ff31f3f2293062c96289 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/lock.lock b/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..a0275171ab905aa564dd8148f619b1159a46618e --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.92201201171875, + "l0_per_token_0.001": 32.85639272460938, + "l0_per_token_0.01": 32.2735626953125, + "l0_per_token_0.05": 29.836496728515623, + "l0_per_token_0.1": 27.11047705078125, + "l0_per_token_0.5": 13.74981396484375, + "l0_per_token_1.0": 7.090305615234375, + "l1_per_token": 25.680720818849334, + "r2_score": 0.8632641941767533, + "l2_loss": 0.09103473252801196 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/config.json b/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..43913a5027e3e99b474c9dfef9490cb3c811038a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/lock.lock b/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..772129e449647115f166f5f9b5eb8e6f15f81235 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.3768498046875, + "l0_per_token_0.001": 32.3122470703125, + "l0_per_token_0.01": 31.737069921875, + "l0_per_token_0.05": 29.33216640625, + "l0_per_token_0.1": 26.64239013671875, + "l0_per_token_0.5": 13.470991357421875, + "l0_per_token_1.0": 6.919353857421875, + "l1_per_token": 25.501685437012753, + "r2_score": 0.8628500154977831, + "l2_loss": 0.09131048067628847 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/code_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x16/l0_64/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..73e84dc57b858de0847bcc4287e6410dc8478333 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 64.986175, + "recon_r2": 0.8943168920266713, + "l1_per_token": 33.00306708557094 + }, + "seed1": { + "l0": 64.34198666666667, + "recon_r2": 0.8927465773695532, + "l1_per_token": 32.64925762786587 + }, + "seed2": { + "l0": 63.99332166666667, + "recon_r2": 0.8931193476001202, + "l1_per_token": 32.57947505524513 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.7338866691739145, + "mcc_abs": 0.7338882563171435, + "act_iou": 0.41278213541041353, + "act_iou_n_pairs": 12288, + "input_r2": 0.9812484291186716, + "matched_mcc_signed": { + "0.6": 0.796904608527718, + "0.7": 0.8513407408239384, + "0.8": 0.8993888481393386, + "0.9": 0.9412500978564922 + }, + "matched_mcc_abs": { + "0.6": 0.7962602508380604, + "0.7": 0.8511200273647856, + "0.8": 0.8993916065573769, + "0.9": 0.941252734986667 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7342046090210208, + "mcc_abs": 0.7342074391438579, + "act_iou": 0.4132143596997026, + "act_iou_n_pairs": 12288, + "input_r2": 0.981774141937448, + "matched_mcc_signed": { + "0.6": 0.7975172279311756, + "0.7": 0.8537603176038489, + "0.8": 0.8997426612439183, + "0.9": 0.941429721654041 + }, + "matched_mcc_abs": { + "0.6": 0.7977557529759814, + "0.7": 0.8535037476764907, + "0.8": 0.8997706353538416, + "0.9": 0.9413834234413334 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7365759792393801, + "mcc_abs": 0.7365787476286085, + "act_iou": 0.4161181369623024, + "act_iou_n_pairs": 12288, + "input_r2": 0.9815256044478343, + "matched_mcc_signed": { + "0.6": 0.7970678881870807, + "0.7": 0.8559156069454162, + "0.8": 0.9000211757452529, + "0.9": 0.9414489258075384 + }, + "matched_mcc_abs": { + "0.6": 0.796593487956282, + "0.7": 0.8553729582793567, + "0.8": 0.8998215333756711, + "0.9": 0.9413314003170167 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 64.44049444444445, + "l0_std": 0.5037035557645152, + "recon_r2_mean": 0.8933942723321149, + "recon_r2_std": 0.0008204631233557927, + "l1_per_token_mean": 32.743933256227315, + "l1_per_token_std": 0.22711265485020962, + "input_r2_mean": 0.9815160585013181, + "input_r2_std": 0.00026298637945269864, + "mcc_signed_mean": 0.7348890858114384, + "mcc_signed_std": 0.0014695164219970418, + "mcc_abs_mean": 0.73489148102987, + "mcc_abs_std": 0.0014699050481685684, + "matched_mcc_signed_per_threshold": { + "0.6": 0.7971632415486581, + "0.7": 0.8536722217910678, + "0.8": 0.8997175617095032, + "0.9": 0.9413762484393572 + }, + "matched_mcc_signed_band": 0.8729823183721466, + "matched_mcc_abs_per_threshold": { + "0.6": 0.796869830590108, + "0.7": 0.853332244440211, + "0.8": 0.8996612584289632, + "0.9": 0.9413225195816723 + }, + "matched_mcc_abs_band": 0.8727964632602386, + "act_iou_mean": 0.4140382106908062, + "act_iou_std": 0.0018141869887660263, + "act_iou_n_pairs": 12288.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/dict_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x16/l0_64/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..2208fbe1985b686d0a83e5cdccf0d52d72c22432 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/dict_metrics_step_final.json @@ -0,0 +1,595 @@ +{ + "schema": 3, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 12288, + "d_in": 768, + "l0": 65, + "l0_raw": 64.96181616210937, + "recon_r2": 0.8948267774733871, + "l1_per_token": 32.91216362071346, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 18, + "dense_rate": 0.00146484375, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 6270, + "0.6": 2923, + "0.7": 994, + "0.8": 235, + "0.9": 16 + }, + "n_dup_clusters_signed": { + "0.5": 2649, + "0.6": 1286, + "0.7": 453, + "0.8": 106, + "0.9": 4 + }, + "n_dup_removed_signed": { + "0.5": 3621, + "0.6": 1637, + "0.7": 541, + "0.8": 129, + "0.9": 12 + }, + "largest_dup_signed": { + "0.5": 10, + "0.6": 10, + "0.7": 9, + "0.8": 9, + "0.9": 9 + }, + "babel_mu1_signed": 0.5482771992683411, + "n_dup_abs": { + "0.5": 6358, + "0.6": 2945, + "0.7": 997, + "0.8": 236, + "0.9": 16 + }, + "n_dup_clusters_abs": { + "0.5": 2687, + "0.6": 1295, + "0.7": 454, + "0.8": 106, + "0.9": 4 + }, + "n_dup_removed_abs": { + "0.5": 3671, + "0.6": 1650, + "0.7": 543, + "0.8": 130, + "0.9": 12 + }, + "largest_dup_abs": { + "0.5": 11, + "0.6": 10, + "0.7": 9, + "0.8": 9, + "0.9": 9 + }, + "babel_mu1_abs": 0.5503074526786804, + "dup_excludes_dead": true + }, + "seed1": { + "d_sae": 12288, + "d_in": 768, + "l0": 64, + "l0_raw": 64.3407306640625, + "recon_r2": 0.8934011879413943, + "l1_per_token": 32.57836625371896, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 18, + "dense_rate": 0.00146484375, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 6317, + "0.6": 2847, + "0.7": 1033, + "0.8": 241, + "0.9": 13 + }, + "n_dup_clusters_signed": { + "0.5": 2686, + "0.6": 1271, + "0.7": 478, + "0.8": 112, + "0.9": 4 + }, + "n_dup_removed_signed": { + "0.5": 3631, + "0.6": 1576, + "0.7": 555, + "0.8": 129, + "0.9": 9 + }, + "largest_dup_signed": { + "0.5": 8, + "0.6": 7, + "0.7": 6, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_signed": 0.5477641224861145, + "n_dup_abs": { + "0.5": 6385, + "0.6": 2862, + "0.7": 1043, + "0.8": 241, + "0.9": 13 + }, + "n_dup_clusters_abs": { + "0.5": 2717, + "0.6": 1277, + "0.7": 482, + "0.8": 112, + "0.9": 4 + }, + "n_dup_removed_abs": { + "0.5": 3668, + "0.6": 1585, + "0.7": 561, + "0.8": 129, + "0.9": 9 + }, + "largest_dup_abs": { + "0.5": 10, + "0.6": 8, + "0.7": 6, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_abs": 0.5498113632202148, + "dup_excludes_dead": true + }, + "seed2": { + "d_sae": 12288, + "d_in": 768, + "l0": 64, + "l0_raw": 64.02126567382813, + "recon_r2": 0.9281115206936617, + "l1_per_token": 32.51113661606752, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 20, + "dense_rate": 0.0016276041666666667, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 6237, + "0.6": 2885, + "0.7": 1034, + "0.8": 240, + "0.9": 14 + }, + "n_dup_clusters_signed": { + "0.5": 2615, + "0.6": 1270, + "0.7": 471, + "0.8": 108, + "0.9": 4 + }, + "n_dup_removed_signed": { + "0.5": 3622, + "0.6": 1615, + "0.7": 563, + "0.8": 132, + "0.9": 10 + }, + "largest_dup_signed": { + "0.5": 14, + "0.6": 8, + "0.7": 8, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_signed": 0.5485987067222595, + "n_dup_abs": { + "0.5": 6315, + "0.6": 2910, + "0.7": 1035, + "0.8": 240, + "0.9": 14 + }, + "n_dup_clusters_abs": { + "0.5": 2652, + "0.6": 1282, + "0.7": 471, + "0.8": 108, + "0.9": 4 + }, + "n_dup_removed_abs": { + "0.5": 3663, + "0.6": 1628, + "0.7": 564, + "0.8": 132, + "0.9": 10 + }, + "largest_dup_abs": { + "0.5": 15, + "0.6": 9, + "0.7": 8, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_abs": 0.5504786372184753, + "dup_excludes_dead": true + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.8539987206459045, + "dcs_abs": 0.8541233539581299, + "frac_sign_flipped": 0.003662109375, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.5": 0.8451948761940002, + "0.6": 0.848082423210144, + "0.7": 0.8508911728858948, + "0.8": 0.8529970049858093, + "0.9": 0.8540554642677307 + }, + "unique_frac_signed": { + "0.5": 0.7049153645833334, + "0.6": 0.8692626953125, + "0.7": 0.9554036458333334, + "0.8": 0.989501953125, + "0.9": 0.9991455078125 + }, + "dedup_dcs_signed": { + "0.5": 0.595790854296259, + "0.6": 0.7372064130468061, + "0.7": 0.812944528782585, + "0.8": 0.8440422024432337, + "0.9": 0.8533256805458223 + }, + "matched_frac_signed": { + "0.5": 0.6416829427083334, + "0.6": 0.749267578125, + "0.7": 0.763427734375, + "0.8": 0.7103678385416666, + "0.9": 0.6044108072916666 + }, + "matched_cos_abs": { + "0.5": 0.8461732268333435, + "0.6": 0.8483722805976868, + "0.7": 0.8511419892311096, + "0.8": 0.8531435132026672, + "0.9": 0.8541707992553711 + }, + "unique_frac_abs": { + "0.5": 0.7013753255208334, + "0.6": 0.8683675130208334, + "0.7": 0.955078125, + "0.8": 0.9894612630208334, + "0.9": 0.9991455078125 + }, + "dedup_dcs_abs": { + "0.5": 0.5934850224172503, + "0.6": 0.7366989274184259, + "0.7": 0.8129070951836184, + "0.8": 0.8441524581115422, + "0.9": 0.8534409169806167 + }, + "matched_frac_abs": { + "0.5": 0.639404296875, + "0.6": 0.7496744791666666, + "0.7": 0.7631022135416666, + "0.8": 0.7106119791666666, + "0.9": 0.6044921875 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.8538036942481995, + "dcs_abs": 0.8539653420448303, + "frac_sign_flipped": 0.004069010416666667, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.5": 0.8456183075904846, + "0.6": 0.8465139865875244, + "0.7": 0.8508668541908264, + "0.8": 0.8527074456214905, + "0.9": 0.8537842035293579 + }, + "unique_frac_signed": { + "0.5": 0.7052815755208334, + "0.6": 0.86767578125, + "0.7": 0.955078125, + "0.8": 0.9893798828125, + "0.9": 0.9991048177083334 + }, + "dedup_dcs_signed": { + "0.5": 0.5963990122666777, + "0.6": 0.7344996846513823, + "0.7": 0.8126443197252229, + "0.8": 0.8436515926223365, + "0.9": 0.8530199110294537 + }, + "matched_frac_signed": { + "0.5": 0.6484375, + "0.6": 0.7472330729166666, + "0.7": 0.75927734375, + "0.8": 0.710205078125, + "0.9": 0.604248046875 + }, + "matched_cos_abs": { + "0.5": 0.8471242785453796, + "0.6": 0.8468260169029236, + "0.7": 0.8509522676467896, + "0.8": 0.8528214693069458, + "0.9": 0.8539299368858337 + }, + "unique_frac_abs": { + "0.5": 0.7015787760416666, + "0.6": 0.8666178385416666, + "0.7": 0.9549560546875, + "0.8": 0.9893391927083334, + "0.9": 0.9991048177083334 + }, + "dedup_dcs_abs": { + "0.5": 0.5943244144970473, + "0.6": 0.7338745323892605, + "0.7": 0.8126220202393597, + "0.8": 0.8437297039684685, + "0.9": 0.8531655139280095 + }, + "matched_frac_abs": { + "0.5": 0.6446940104166666, + "0.6": 0.7465006510416666, + "0.7": 0.7594401041666666, + "0.8": 0.70947265625, + "0.9": 0.6045735677083334 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.8555665612220764, + "dcs_abs": 0.8557350635528564, + "frac_sign_flipped": 0.0038248697916666665, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 12288, + "n_alive_b": 12288, + "matched_cos_signed": { + "0.5": 0.846964418888092, + "0.6": 0.8478524088859558, + "0.7": 0.8524401783943176, + "0.8": 0.8543527126312256, + "0.9": 0.8555065393447876 + }, + "unique_frac_signed": { + "0.5": 0.7048746744791666, + "0.6": 0.8701578776041666, + "0.7": 0.9545084635416666, + "0.8": 0.9893798828125, + "0.9": 0.9992268880208334 + }, + "dedup_dcs_signed": { + "0.5": 0.5970037690591804, + "0.6": 0.7377654526377834, + "0.7": 0.8136613649403444, + "0.8": 0.8452793867036235, + "0.9": 0.8548451369909648 + }, + "matched_frac_signed": { + "0.5": 0.6459147135416666, + "0.6": 0.74951171875, + "0.7": 0.7603352864583334, + "0.8": 0.7146809895833334, + "0.9": 0.6106770833333334 + }, + "matched_cos_abs": { + "0.5": 0.847797155380249, + "0.6": 0.8481495976448059, + "0.7": 0.8524317741394043, + "0.8": 0.8545071482658386, + "0.9": 0.8556612133979797 + }, + "unique_frac_abs": { + "0.5": 0.7017008463541666, + "0.6": 0.8692626953125, + "0.7": 0.9542236328125, + "0.8": 0.9893798828125, + "0.9": 0.9992268880208334 + }, + "dedup_dcs_abs": { + "0.5": 0.5948999814669756, + "0.6": 0.7372648052769364, + "0.7": 0.8134105442441069, + "0.8": 0.845432182213699, + "0.9": 0.8549996914637935 + }, + "matched_frac_abs": { + "0.5": 0.6427408854166666, + "0.6": 0.7493489583333334, + "0.7": 0.760498046875, + "0.8": 0.7150065104166666, + "0.9": 0.611083984375 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 19, + "n_dense_std": 1.1547005383792515, + "l0_mean": 64, + "l0_std": 0.47826775918859304, + "recon_r2_mean": 0.905446495369481, + "recon_r2_std": 0.019641425765146055, + "l1_per_token_mean": 32.66722216349998, + "l1_per_token_std": 0.21477242407041655, + "babel_mu1_signed_mean": 0.5482133428255717, + "babel_mu1_signed_std": 0.00042094054897076573, + "n_dup_signed_per_threshold": { + "0.5": 6274.666666666667, + "0.6": 2885.0, + "0.7": 1020.3333333333334, + "0.8": 238.66666666666666, + "0.9": 14.333333333333334 + }, + "n_dup_signed_band": 1039.5833333333333, + "n_dup_signed_band_std": 5.305264680799001, + "n_dup_clusters_signed_per_threshold": { + "0.5": 2650.0, + "0.6": 1275.6666666666667, + "0.7": 467.3333333333333, + "0.8": 108.66666666666667, + "0.9": 4.0 + }, + "n_dup_clusters_signed_band": 463.9166666666667, + "n_dup_clusters_signed_band_std": 2.0816659994661326, + "largest_dup_signed_per_threshold": { + "0.5": 10.666666666666666, + "0.6": 8.333333333333334, + "0.7": 7.666666666666667, + "0.8": 7.333333333333333, + "0.9": 7.333333333333333 + }, + "largest_dup_signed_band": 7.666666666666666, + "largest_dup_signed_band_std": 1.5069284433354269, + "dcs_signed_mean": 0.8544563253720602, + "dcs_signed_std": 0.0009664246250724767, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.5963978785407057, + "0.6": 0.7364905167786572, + "0.7": 0.8130834044827174, + "0.8": 0.8443243939230646, + "0.9": 0.8537302428554137 + }, + "dedup_dcs_signed_band": 0.8119071395099632, + "dedup_dcs_signed_band_std": 0.0009672709687211982, + "matched_cos_signed_per_threshold": { + "0.5": 0.8459258675575256, + "0.6": 0.8474829395612081, + "0.7": 0.8513994018236796, + "0.8": 0.8533523877461752, + "0.9": 0.8544487357139587 + }, + "matched_cos_signed_band": 0.8516708662112554, + "matched_cos_signed_band_std": 0.0007977189184384786, + "unique_frac_signed_per_threshold": { + "0.5": 0.7050238715277778, + "0.6": 0.8690321180555555, + "0.7": 0.9549967447916666, + "0.8": 0.9894205729166666, + "0.9": 0.9991590711805557 + }, + "unique_frac_signed_band": 0.9531521267361112, + "unique_frac_signed_band_std": 0.0002966356967507547, + "matched_frac_signed_per_threshold": { + "0.5": 0.6453450520833334, + "0.6": 0.7486707899305555, + "0.7": 0.7610134548611112, + "0.8": 0.7117513020833334, + "0.9": 0.6064453125 + }, + "matched_frac_signed_band": 0.70697021484375, + "matched_frac_signed_band_std": 0.0017823705513444, + "babel_mu1_abs_mean": 0.5501991510391235, + "babel_mu1_abs_std": 0.0003465697347384359, + "n_dup_abs_per_threshold": { + "0.5": 6352.666666666667, + "0.6": 2905.6666666666665, + "0.7": 1025.0, + "0.8": 239.0, + "0.9": 14.333333333333334 + }, + "n_dup_abs_band": 1045.9999999999998, + "n_dup_abs_band_std": 5.448623679425842, + "n_dup_clusters_abs_per_threshold": { + "0.5": 2685.3333333333335, + "0.6": 1284.6666666666667, + "0.7": 469.0, + "0.8": 108.66666666666667, + "0.9": 4.0 + }, + "n_dup_clusters_abs_band": 466.58333333333337, + "n_dup_clusters_abs_band_std": 2.0207259421636903, + "largest_dup_abs_per_threshold": { + "0.5": 12.0, + "0.6": 9.0, + "0.7": 7.666666666666667, + "0.8": 7.333333333333333, + "0.9": 7.333333333333333 + }, + "largest_dup_abs_band": 7.833333333333333, + "largest_dup_abs_band_std": 1.3768926368215255, + "dcs_abs_mean": 0.8546079198519388, + "dcs_abs_std": 0.000979327132818626, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.5942364727937578, + "0.6": 0.735946088361541, + "0.7": 0.812979886555695, + "0.8": 0.84443811476457, + "0.9": 0.8538687074574733 + }, + "dedup_dcs_abs_band": 0.8118081992848198, + "dedup_dcs_abs_band_std": 0.0009644586931023847, + "matched_cos_abs_per_threshold": { + "0.5": 0.8470315535863241, + "0.6": 0.8477826317151388, + "0.7": 0.8515086770057678, + "0.8": 0.8534907102584839, + "0.9": 0.8545873165130615 + }, + "matched_cos_abs_band": 0.851842333873113, + "matched_cos_abs_band_std": 0.0007862706013482083, + "unique_frac_abs_per_threshold": { + "0.5": 0.7015516493055555, + "0.6": 0.8680826822916666, + "0.7": 0.9547526041666666, + "0.8": 0.9893934461805557, + "0.9": 0.9991590711805557 + }, + "unique_frac_abs_band": 0.9528469509548612, + "unique_frac_abs_band_std": 0.00029663569675078576, + "matched_frac_abs_per_threshold": { + "0.5": 0.6422797309027777, + "0.6": 0.7485080295138888, + "0.7": 0.761013454861111, + "0.8": 0.711697048611111, + "0.9": 0.6067165798611112 + }, + "matched_frac_abs_band": 0.7069837782118056, + "matched_frac_abs_band_std": 0.0019938497042951775, + "alive_source": "stats", + "frac_denominator": "d_sae", + "d_sae": 12288 + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/config.json b/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5949eb30b783de244a80f491fa25f7af3f2edec1 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/lock.lock b/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..b13fd4713c61486e8c3a1f25b8bb173336468db6 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.96181616210937, + "l0_per_token_0.001": 64.76177797851562, + "l0_per_token_0.01": 62.99214208984375, + "l0_per_token_0.05": 55.7901041015625, + "l0_per_token_0.1": 48.13519321289063, + "l0_per_token_0.5": 17.810970654296874, + "l0_per_token_1.0": 7.60318251953125, + "l1_per_token": 32.91216362071346, + "r2_score": 0.8948267774733871, + "l2_loss": 0.07002128026506657 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/config.json b/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..4c15515c994745470445ff31f3f2293062c96289 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/lock.lock b/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..f75e44e46e0550f9a4db38183b21059b2e6a11e3 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.3407306640625, + "l0_per_token_0.001": 64.14350966796874, + "l0_per_token_0.01": 62.398555810546874, + "l0_per_token_0.05": 55.2943744140625, + "l0_per_token_0.1": 47.73980517578125, + "l0_per_token_0.5": 17.746207177734377, + "l0_per_token_1.0": 7.60479794921875, + "l1_per_token": 32.57836625371896, + "r2_score": 0.8934011879413943, + "l2_loss": 0.07097039641616058 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/config.json b/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..43913a5027e3e99b474c9dfef9490cb3c811038a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 16, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/lock.lock b/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/lock.lock new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..46579fb83ccbfa211f0b2b90a9185798e5cafbe1 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.02126567382813, + "l0_per_token_0.001": 63.82433764648437, + "l0_per_token_0.01": 62.083716748046875, + "l0_per_token_0.05": 54.9993046875, + "l0_per_token_0.1": 47.46670239257813, + "l0_per_token_0.5": 17.61234931640625, + "l0_per_token_1.0": 7.525110107421875, + "l1_per_token": 32.51113661606752, + "r2_score": 0.9281115206936617, + "l2_loss": 0.07077603618059644 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..4ac90e1883589da6c300eefe3cbf3fd5d4044d4c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 12288, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/code_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x4/l0_128/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..fa7811db4e30205cd67b5c2dca4efb07250d883a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 125.429215, + "recon_r2": 0.9040893949955573, + "l1_per_token": 48.96858258553746 + }, + "seed1": { + "l0": 127.41857166666667, + "recon_r2": 0.9047144349381544, + "l1_per_token": 49.547137375215314 + }, + "seed2": { + "l0": 126.22322333333334, + "recon_r2": 0.9043200820072059, + "l1_per_token": 49.07499270022832 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.7793609190206778, + "mcc_abs": 0.7793717400300944, + "act_iou": 0.4899197592825965, + "act_iou_n_pairs": 3072, + "input_r2": 0.9834704700687155, + "matched_mcc_signed": { + "0.6": 0.8762186580968883, + "0.7": 0.9103201026265718, + "0.8": 0.9346459920269303, + "0.9": 0.9594665426542683 + }, + "matched_mcc_abs": { + "0.6": 0.8747610999991104, + "0.7": 0.9088798563338277, + "0.8": 0.9338134363174554, + "0.9": 0.9597518303340247 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7747081471495839, + "mcc_abs": 0.774715398588029, + "act_iou": 0.4868974865840639, + "act_iou_n_pairs": 3072, + "input_r2": 0.9837281997593362, + "matched_mcc_signed": { + "0.6": 0.8776902470558942, + "0.7": 0.9084543752762022, + "0.8": 0.9343068779527339, + "0.9": 0.960338320369024 + }, + "matched_mcc_abs": { + "0.6": 0.8749769332053454, + "0.7": 0.9053769919145921, + "0.8": 0.9330049322220012, + "0.9": 0.960273406494499 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7783732922051462, + "mcc_abs": 0.7783785952068426, + "act_iou": 0.4911844192446013, + "act_iou_n_pairs": 3072, + "input_r2": 0.9840436675869002, + "matched_mcc_signed": { + "0.6": 0.8763806116055858, + "0.7": 0.908536065939906, + "0.8": 0.9371528615187292, + "0.9": 0.9607567185074571 + }, + "matched_mcc_abs": { + "0.6": 0.8748024981756566, + "0.7": 0.9059932308396517, + "0.8": 0.9360217230506155, + "0.9": 0.9606726375331229 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 126.35700333333334, + "l0_std": 1.0014029174626846, + "recon_r2_mean": 0.9043746373136393, + "recon_r2_std": 0.0003160711052207556, + "l1_per_token_mean": 49.196904220327035, + "l1_per_token_std": 0.3079419185729345, + "input_r2_mean": 0.9837474458049839, + "input_r2_std": 0.0002870830131130944, + "mcc_signed_mean": 0.7774807861251359, + "mcc_signed_std": 0.0024514275123388987, + "mcc_abs_mean": 0.7774885779416553, + "mcc_abs_std": 0.002452443058568846, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8767631722527894, + "0.7": 0.9091035146142268, + "0.8": 0.9353685771661312, + "0.9": 0.9601871938435832 + }, + "matched_mcc_signed_band": 0.9203556144691827, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8748468437933709, + "0.7": 0.9067500263626904, + "0.8": 0.9342800305300241, + "0.9": 0.9602326247872156 + }, + "matched_mcc_abs_band": 0.9190273813683253, + "act_iou_mean": 0.4893338883704206, + "act_iou_std": 0.002202698675101979, + "act_iou_n_pairs": 3072.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/dict_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x4/l0_128/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..e2a7bf0638590feb05bc74412932da33e5dd6b27 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 3072, + "d_in": 768, + "l0": 125, + "l0_raw": 125.32729682617187, + "recon_r2": 0.904465276945699, + "l1_per_token": 48.84986610206857, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 178, + "dense_rate": 0.057942708333333336, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 343, + "0.6": 108, + "0.7": 35, + "0.8": 16, + "0.9": 8 + }, + "n_dup_clusters_signed": { + "0.5": 153, + "0.6": 48, + "0.7": 14, + "0.8": 5, + "0.9": 2 + }, + "largest_dup_signed": { + "0.5": 8, + "0.6": 7, + "0.7": 6, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_signed": 0.39574432373046875, + "n_dup_abs": { + "0.5": 478, + "0.6": 179, + "0.7": 87, + "0.8": 58, + "0.9": 18 + }, + "n_dup_clusters_abs": { + "0.5": 221, + "0.6": 83, + "0.7": 39, + "0.8": 26, + "0.9": 7 + }, + "largest_dup_abs": { + "0.5": 7, + "0.6": 7, + "0.7": 7, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_abs": 0.4202030897140503 + }, + "seed1": { + "d_sae": 3072, + "d_in": 768, + "l0": 127, + "l0_raw": 127.35413041992187, + "recon_r2": 0.9051924332629038, + "l1_per_token": 49.430623173551865, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 186, + "dense_rate": 0.060546875, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 348, + "0.6": 123, + "0.7": 44, + "0.8": 17, + "0.9": 9 + }, + "n_dup_clusters_signed": { + "0.5": 155, + "0.6": 56, + "0.7": 19, + "0.8": 6, + "0.9": 3 + }, + "largest_dup_signed": { + "0.5": 6, + "0.6": 6, + "0.7": 5, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_signed": 0.39512038230895996, + "n_dup_abs": { + "0.5": 460, + "0.6": 184, + "0.7": 93, + "0.8": 53, + "0.9": 19 + }, + "n_dup_clusters_abs": { + "0.5": 209, + "0.6": 85, + "0.7": 43, + "0.8": 24, + "0.9": 8 + }, + "largest_dup_abs": { + "0.5": 6, + "0.6": 6, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_abs": 0.4189646244049072 + }, + "seed2": { + "d_sae": 3072, + "d_in": 768, + "l0": 126, + "l0_raw": 126.14133334960937, + "recon_r2": 0.9047218867286111, + "l1_per_token": 48.95884410146535, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 190, + "dense_rate": 0.061848958333333336, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 392, + "0.6": 119, + "0.7": 38, + "0.8": 15, + "0.9": 8 + }, + "n_dup_clusters_signed": { + "0.5": 176, + "0.6": 52, + "0.7": 16, + "0.8": 5, + "0.9": 2 + }, + "largest_dup_signed": { + "0.5": 8, + "0.6": 7, + "0.7": 6, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_signed": 0.3975372314453125, + "n_dup_abs": { + "0.5": 527, + "0.6": 195, + "0.7": 98, + "0.8": 59, + "0.9": 18 + }, + "n_dup_clusters_abs": { + "0.5": 241, + "0.6": 90, + "0.7": 45, + "0.8": 27, + "0.9": 7 + }, + "largest_dup_abs": { + "0.5": 7, + "0.6": 7, + "0.7": 7, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_abs": 0.42296162247657776 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.8418418765068054, + "dcs_abs": 0.8425191044807434, + "frac_sign_flipped": 0.0166015625, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.8362765312194824, + "0.6": 0.8406468033790588, + "0.7": 0.8418320417404175, + "0.8": 0.8415083289146423, + "0.9": 0.8416629433631897 + }, + "unique_frac_signed": { + "0.5": 0.9376627604166666, + "0.6": 0.9793294270833334, + "0.7": 0.9925130208333334, + "0.8": 0.9964192708333334, + "0.9": 0.998046875 + }, + "dedup_dcs_signed": { + "0.5": 0.7841453607349346, + "0.6": 0.8232701523326493, + "0.7": 0.8355292627820745, + "0.8": 0.8384951154973048, + "0.9": 0.8400190704269335 + }, + "matched_frac_signed": { + "0.5": 0.8372395833333334, + "0.6": 0.8225911458333334, + "0.7": 0.7747395833333334, + "0.8": 0.7180989583333334, + "0.9": 0.6197916666666666 + }, + "matched_cos_abs": { + "0.5": 0.8415570855140686, + "0.6": 0.8432239294052124, + "0.7": 0.8444792032241821, + "0.8": 0.8440569639205933, + "0.9": 0.8426223993301392 + }, + "unique_frac_abs": { + "0.5": 0.9173177083333334, + "0.6": 0.96826171875, + "0.7": 0.9840494791666666, + "0.8": 0.9900716145833334, + "0.9": 0.9964192708333334 + }, + "dedup_dcs_abs": { + "0.5": 0.7719752171154445, + "0.6": 0.8164614511770196, + "0.7": 0.8310093200998381, + "0.8": 0.8356768410691682, + "0.9": 0.8396051967283711 + }, + "matched_frac_abs": { + "0.5": 0.8232421875, + "0.6": 0.8180338541666666, + "0.7": 0.7724609375, + "0.8": 0.716796875, + "0.9": 0.6184895833333334 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.8373012542724609, + "dcs_abs": 0.8380201458930969, + "frac_sign_flipped": 0.0146484375, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.8336585164070129, + "0.6": 0.8356645107269287, + "0.7": 0.8361518383026123, + "0.8": 0.836951732635498, + "0.9": 0.836991012096405 + }, + "unique_frac_signed": { + "0.5": 0.9339192708333334, + "0.6": 0.9793294270833334, + "0.7": 0.9930013020833334, + "0.8": 0.99658203125, + "0.9": 0.998046875 + }, + "dedup_dcs_signed": { + "0.5": 0.778569753766836, + "0.6": 0.8183908465240772, + "0.7": 0.8302998641738668, + "0.8": 0.8340910577680916, + "0.9": 0.8353562640259042 + }, + "matched_frac_signed": { + "0.5": 0.8274739583333334, + "0.6": 0.8111979166666666, + "0.7": 0.7685546875, + "0.8": 0.7096354166666666, + "0.9": 0.6110026041666666 + }, + "matched_cos_abs": { + "0.5": 0.8397945165634155, + "0.6": 0.8395894765853882, + "0.7": 0.8402825593948364, + "0.8": 0.8397801518440247, + "0.9": 0.8383629322052002 + }, + "unique_frac_abs": { + "0.5": 0.91162109375, + "0.6": 0.96728515625, + "0.7": 0.9835611979166666, + "0.8": 0.9895833333333334, + "0.9": 0.9964192708333334 + }, + "dedup_dcs_abs": { + "0.5": 0.7655743957147934, + "0.6": 0.8121224380447529, + "0.7": 0.8264693207068678, + "0.8": 0.8310324419289827, + "0.9": 0.8353609816016009 + }, + "matched_frac_abs": { + "0.5": 0.8098958333333334, + "0.6": 0.8089192708333334, + "0.7": 0.7698567708333334, + "0.8": 0.7099609375, + "0.9": 0.6110026041666666 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.8415473103523254, + "dcs_abs": 0.8422792553901672, + "frac_sign_flipped": 0.015625, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.8374631404876709, + "0.6": 0.8392462730407715, + "0.7": 0.8409841656684875, + "0.8": 0.8414010405540466, + "0.9": 0.8413373231887817 + }, + "unique_frac_signed": { + "0.5": 0.9334309895833334, + "0.6": 0.9781901041666666, + "0.7": 0.9923502604166666, + "0.8": 0.99658203125, + "0.9": 0.998046875 + }, + "dedup_dcs_signed": { + "0.5": 0.7817140479649728, + "0.6": 0.820942399247239, + "0.7": 0.8345508558074167, + "0.8": 0.8385251580912154, + "0.9": 0.8396940862294286 + }, + "matched_frac_signed": { + "0.5": 0.83203125, + "0.6": 0.8193359375, + "0.7": 0.7740885416666666, + "0.8": 0.7115885416666666, + "0.9": 0.6201171875 + }, + "matched_cos_abs": { + "0.5": 0.8448786735534668, + "0.6": 0.8437333703041077, + "0.7": 0.8442956805229187, + "0.8": 0.8442563414573669, + "0.9": 0.8425873517990112 + }, + "unique_frac_abs": { + "0.5": 0.91259765625, + "0.6": 0.966796875, + "0.7": 0.9832356770833334, + "0.8": 0.9900716145833334, + "0.9": 0.9964192708333334 + }, + "dedup_dcs_abs": { + "0.5": 0.7710342973005027, + "0.6": 0.8157187857432291, + "0.7": 0.8301416350974857, + "0.8": 0.8358742391089133, + "0.9": 0.8395702746929601 + }, + "matched_frac_abs": { + "0.5": 0.8177083333333334, + "0.6": 0.8157552083333334, + "0.7": 0.7734375, + "0.8": 0.7109375, + "0.9": 0.6204427083333334 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 185, + "n_dense_std": 6.110100926607786, + "l0_mean": 126, + "l0_std": 1.0199335445673514, + "recon_r2_mean": 0.9047931989790712, + "recon_r2_std": null, + "l1_per_token_mean": 49.07977779236193, + "l1_per_token_std": 0.3086882118319704, + "babel_mu1_signed_mean": 0.3961339791615804, + "babel_mu1_signed_std": 0.0012546567073180258, + "n_dup_signed_per_threshold": { + "0.5": 361.0, + "0.6": 116.66666666666667, + "0.7": 39.0, + "0.8": 16.0, + "0.9": 8.333333333333334 + }, + "n_dup_signed_band": 45.00000000000001, + "n_dup_signed_band_std": 3.25, + "n_dup_clusters_signed_per_threshold": { + "0.5": 161.33333333333334, + "0.6": 52.0, + "0.7": 16.333333333333332, + "0.8": 5.333333333333333, + "0.9": 2.3333333333333335 + }, + "n_dup_clusters_signed_band": 18.999999999999996, + "n_dup_clusters_signed_band_std": 1.8874586088176875, + "largest_dup_signed_per_threshold": { + "0.5": 7.333333333333333, + "0.6": 6.666666666666667, + "0.7": 5.666666666666667, + "0.8": 5.666666666666667, + "0.9": 5.666666666666667 + }, + "largest_dup_signed_band": 5.916666666666667, + "largest_dup_signed_band_std": 0.5773502691896258, + "dcs_signed_mean": 0.8402301470438639, + "dcs_signed_std": 0.002540767985201289, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.7814763874889145, + "0.6": 0.8208677993679885, + "0.7": 0.8334599942544526, + "0.8": 0.837037110452204, + "0.9": 0.8383564735607555 + }, + "dedup_dcs_signed_band": 0.8324303444088501, + "dedup_dcs_signed_band_std": 0.002547945211739332, + "matched_cos_signed_per_threshold": { + "0.5": 0.8357993960380554, + "0.6": 0.8385191957155863, + "0.7": 0.8396560152371725, + "0.8": 0.8399537007013956, + "0.9": 0.8399970928827921 + }, + "matched_cos_signed_band": 0.8395315011342366, + "matched_cos_signed_band_std": 0.002698410690292952, + "unique_frac_signed_per_threshold": { + "0.5": 0.9350043402777778, + "0.6": 0.9789496527777778, + "0.7": 0.9926215277777778, + "0.8": 0.9965277777777778, + "0.9": 0.998046875 + }, + "unique_frac_signed_band": 0.9915364583333334, + "unique_frac_signed_band_std": 0.00022655291189908116, + "matched_frac_signed_per_threshold": { + "0.5": 0.832248263888889, + "0.6": 0.8177083333333334, + "0.7": 0.7724609375, + "0.8": 0.7131076388888888, + "0.9": 0.616970486111111 + }, + "matched_frac_signed_band": 0.7300618489583334, + "matched_frac_signed_band_std": 0.004480348179850558, + "babel_mu1_abs_mean": 0.4207097788651784, + "babel_mu1_abs_std": 0.002046105769049513, + "n_dup_abs_per_threshold": { + "0.5": 488.3333333333333, + "0.6": 186.0, + "0.7": 92.66666666666667, + "0.8": 56.666666666666664, + "0.9": 18.333333333333332 + }, + "n_dup_abs_band": 88.41666666666667, + "n_dup_abs_band_std": 3.642915499065732, + "n_dup_clusters_abs_per_threshold": { + "0.5": 223.66666666666666, + "0.6": 86.0, + "0.7": 42.333333333333336, + "0.8": 25.666666666666668, + "0.9": 7.333333333333333 + }, + "n_dup_clusters_abs_band": 40.333333333333336, + "n_dup_clusters_abs_band_std": 1.773649721149397, + "largest_dup_abs_per_threshold": { + "0.5": 6.666666666666667, + "0.6": 6.666666666666667, + "0.7": 6.666666666666667, + "0.8": 5.666666666666667, + "0.9": 5.666666666666667 + }, + "largest_dup_abs_band": 6.166666666666667, + "largest_dup_abs_band_std": 0.5773502691896258, + "dcs_abs_mean": 0.8409395019213358, + "dcs_abs_std": 0.0025310791396702314, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.7695279700435802, + "0.6": 0.8147675583216673, + "0.7": 0.8292067586347306, + "0.8": 0.8341945073690215, + "0.9": 0.8381788176743107 + }, + "dedup_dcs_abs_band": 0.8290869104999324, + "dedup_dcs_abs_band_std": 0.0024666931712424463, + "matched_cos_abs_per_threshold": { + "0.5": 0.8420767585436503, + "0.6": 0.8421822587649027, + "0.7": 0.8430191477139791, + "0.8": 0.842697819073995, + "0.9": 0.8411908944447836 + }, + "matched_cos_abs_band": 0.8422725299994152, + "matched_cos_abs_band_std": 0.002398590784803484, + "unique_frac_abs_per_threshold": { + "0.5": 0.9138454861111112, + "0.6": 0.9674479166666666, + "0.7": 0.9836154513888888, + "0.8": 0.9899088541666666, + "0.9": 0.9964192708333334 + }, + "unique_frac_abs_band": 0.9843478732638888, + "unique_frac_abs_band_std": 0.00030810049619139505, + "matched_frac_abs_per_threshold": { + "0.5": 0.8169487847222223, + "0.6": 0.8142361111111112, + "0.7": 0.7719184027777778, + "0.8": 0.7125651041666666, + "0.9": 0.6166449652777778 + }, + "matched_frac_abs_band": 0.7288411458333333, + "matched_frac_abs_band_std": 0.003444988686282012, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/seed0/config.json b/lm/pythia-160m_layer8/relu/x4/l0_128/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..920481756aae71a20da4a01e54beb062205c045a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/seed0/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..fdc3e8eb1c9b8bbb971d5837e49861daef77c0f8 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 125.32729682617187, + "l0_per_token_0.001": 124.87649985351563, + "l0_per_token_0.01": 120.894287890625, + "l0_per_token_0.05": 104.78733701171875, + "l0_per_token_0.1": 87.90249008789063, + "l0_per_token_0.5": 26.1540837890625, + "l0_per_token_1.0": 9.663599365234376, + "l1_per_token": 48.84986610206857, + "r2_score": 0.904465276945699, + "l2_loss": 0.06360424694924642 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/seed1/config.json b/lm/pythia-160m_layer8/relu/x4/l0_128/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2b729114fa7a39fd6d4550e7ebd1e771051ed38a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/seed1/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..0ec9f1c5c469466a877463c7607bd921a15abbc2 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 127.35413041992187, + "l0_per_token_0.001": 126.89603793945312, + "l0_per_token_0.01": 122.85050083007812, + "l0_per_token_0.05": 106.47781967773437, + "l0_per_token_0.1": 89.30079716796875, + "l0_per_token_0.5": 26.475064599609375, + "l0_per_token_1.0": 9.765594140625, + "l1_per_token": 49.430623173551865, + "r2_score": 0.9051924332629038, + "l2_loss": 0.06312012736956321 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/seed2/config.json b/lm/pythia-160m_layer8/relu/x4/l0_128/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d2ac614f301c676dc8383e32f921621c96254746 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/seed2/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..5420297a19450ff433a8b7de4ca822d718d51333 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 126.14133334960937, + "l0_per_token_0.001": 125.68483525390624, + "l0_per_token_0.01": 121.65090620117188, + "l0_per_token_0.05": 105.3390576171875, + "l0_per_token_0.1": 88.24616020507813, + "l0_per_token_0.5": 25.99233037109375, + "l0_per_token_1.0": 9.564248828125, + "l1_per_token": 48.95884410146535, + "r2_score": 0.9047218867286111, + "l2_loss": 0.06343340360056518 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/code_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x4/l0_32/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..4f0fafc91a304b44a94543fd68483c05b3d3edac --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 31.54979, + "recon_r2": 0.8443332316675541, + "l1_per_token": 26.827647675370176 + }, + "seed1": { + "l0": 31.147358333333333, + "recon_r2": 0.8435668515077319, + "l1_per_token": 26.689876319568704 + }, + "seed2": { + "l0": 31.024706666666667, + "recon_r2": 0.8437621180990215, + "l1_per_token": 26.68355323391234 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.8143934755886951, + "mcc_abs": 0.8144099273588564, + "act_iou": 0.5312589213269578, + "act_iou_n_pairs": 3072, + "input_r2": 0.983856178266369, + "matched_mcc_signed": { + "0.6": 0.8707208099009863, + "0.7": 0.8959978463503047, + "0.8": 0.924041133117871, + "0.9": 0.9525620582343736 + }, + "matched_mcc_abs": { + "0.6": 0.8688136122272087, + "0.7": 0.8951547306633683, + "0.8": 0.9239676444119731, + "0.9": 0.9521308181462536 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.8216520949170466, + "mcc_abs": 0.8216585827685664, + "act_iou": 0.5395625550152507, + "act_iou_n_pairs": 3072, + "input_r2": 0.9844956323757761, + "matched_mcc_signed": { + "0.6": 0.8742908756531151, + "0.7": 0.9035358124785356, + "0.8": 0.9244610452516685, + "0.9": 0.9497572448513394 + }, + "matched_mcc_abs": { + "0.6": 0.872368051493248, + "0.7": 0.9029393775059511, + "0.8": 0.9240329047528829, + "0.9": 0.9498549278339823 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.824710440376521, + "mcc_abs": 0.8247238370436433, + "act_iou": 0.5436213267449714, + "act_iou_n_pairs": 3072, + "input_r2": 0.9844074229852755, + "matched_mcc_signed": { + "0.6": 0.8798602293707486, + "0.7": 0.9002395014410597, + "0.8": 0.9248024235521862, + "0.9": 0.9530430513704823 + }, + "matched_mcc_abs": { + "0.6": 0.8785081479747772, + "0.7": 0.8999606236108613, + "0.8": 0.9243159645583603, + "0.9": 0.9529596634950683 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 31.240618333333334, + "l0_std": 0.27468381356773025, + "recon_r2_mean": 0.8438874004247693, + "recon_r2_std": 0.000398254169094123, + "l1_per_token_mean": 26.733692409617074, + "l1_per_token_std": 0.08142904472672315, + "input_r2_mean": 0.984253077875807, + "input_r2_std": 0.00034654321393166296, + "mcc_signed_mean": 0.8202520036274209, + "mcc_signed_std": 0.005299069005285166, + "mcc_abs_mean": 0.8202641157236887, + "mcc_abs_std": 0.005296469295759436, + "matched_mcc_signed_per_threshold": { + "0.6": 0.87495730497495, + "0.7": 0.8999243867566333, + "0.8": 0.9244348673072419, + "0.9": 0.9517874514853983 + }, + "matched_mcc_signed_band": 0.9127760026310558, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8732299372317446, + "0.7": 0.8993515772600601, + "0.8": 0.9241055045744054, + "0.9": 0.9516484698251014 + }, + "matched_mcc_abs_band": 0.9120838722228279, + "act_iou_mean": 0.53814760102906, + "act_iou_std": 0.006301494903363068, + "act_iou_n_pairs": 3072.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/dict_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x4/l0_32/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..02a9c8cd25783fbaa35b6f53958f34c24a0ddc9b --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 3072, + "d_in": 768, + "l0": 32, + "l0_raw": 31.55881953125, + "recon_r2": 0.8450262789430625, + "l1_per_token": 26.77954580725691, + "stats_source": "stats", + "n_dead": 2, + "dead_rate": 0.0006510416666666666, + "alive_source": "stats", + "n_dense": 12, + "dense_rate": 0.00390625, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1555, + "0.6": 679, + "0.7": 242, + "0.8": 81, + "0.9": 10 + }, + "n_dup_clusters_signed": { + "0.5": 666, + "0.6": 302, + "0.7": 111, + "0.8": 38, + "0.9": 3 + }, + "largest_dup_signed": { + "0.5": 8, + "0.6": 6, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_signed": 0.5471508502960205, + "n_dup_abs": { + "0.5": 1568, + "0.6": 694, + "0.7": 246, + "0.8": 83, + "0.9": 10 + }, + "n_dup_clusters_abs": { + "0.5": 671, + "0.6": 308, + "0.7": 113, + "0.8": 39, + "0.9": 3 + }, + "largest_dup_abs": { + "0.5": 8, + "0.6": 6, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_abs": 0.5490069389343262 + }, + "seed1": { + "d_sae": 3072, + "d_in": 768, + "l0": 31, + "l0_raw": 31.1541205078125, + "recon_r2": 0.8443971747611574, + "l1_per_token": 26.650040147504185, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 11, + "dense_rate": 0.0035807291666666665, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1544, + "0.6": 688, + "0.7": 217, + "0.8": 61, + "0.9": 12 + }, + "n_dup_clusters_signed": { + "0.5": 655, + "0.6": 309, + "0.7": 98, + "0.8": 28, + "0.9": 4 + }, + "largest_dup_signed": { + "0.5": 9, + "0.6": 7, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_signed": 0.5446761250495911, + "n_dup_abs": { + "0.5": 1558, + "0.6": 708, + "0.7": 222, + "0.8": 61, + "0.9": 12 + }, + "n_dup_clusters_abs": { + "0.5": 661, + "0.6": 319, + "0.7": 100, + "0.8": 28, + "0.9": 4 + }, + "largest_dup_abs": { + "0.5": 9, + "0.6": 7, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_abs": 0.5465609431266785 + }, + "seed2": { + "d_sae": 3072, + "d_in": 768, + "l0": 31, + "l0_raw": 31.03178271484375, + "recon_r2": 0.8444554523274143, + "l1_per_token": 26.633970499149342, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 10, + "dense_rate": 0.0032552083333333335, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1530, + "0.6": 671, + "0.7": 231, + "0.8": 63, + "0.9": 8 + }, + "n_dup_clusters_signed": { + "0.5": 655, + "0.6": 301, + "0.7": 108, + "0.8": 30, + "0.9": 3 + }, + "largest_dup_signed": { + "0.5": 9, + "0.6": 6, + "0.7": 5, + "0.8": 4, + "0.9": 4 + }, + "babel_mu1_signed": 0.5434955954551697, + "n_dup_abs": { + "0.5": 1548, + "0.6": 696, + "0.7": 234, + "0.8": 63, + "0.9": 8 + }, + "n_dup_clusters_abs": { + "0.5": 663, + "0.6": 313, + "0.7": 109, + "0.8": 30, + "0.9": 3 + }, + "largest_dup_abs": { + "0.5": 9, + "0.6": 6, + "0.7": 5, + "0.8": 4, + "0.9": 4 + }, + "babel_mu1_abs": 0.5457243919372559 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.8947961926460266, + "dcs_abs": 0.8948999047279358, + "frac_sign_flipped": 0.005208333333333333, + "alive_source": "stats", + "n_alive_a": 3070, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.8949230313301086, + "0.6": 0.89237380027771, + "0.7": 0.8960789442062378, + "0.8": 0.8970603346824646, + "0.9": 0.8949152827262878 + }, + "unique_frac_signed": { + "0.5": 0.7108433734939759, + "0.6": 0.8770758710517746, + "0.7": 0.9594594594594594, + "0.8": 0.9876261803972647, + "0.9": 0.9975577987626181 + }, + "dedup_dcs_signed": { + "0.5": 0.6361501066081495, + "0.6": 0.7826795281823549, + "0.7": 0.85975141944112, + "0.8": 0.8859602719283344, + "0.9": 0.8927297195154618 + }, + "matched_frac_signed": { + "0.5": 0.6600455877564312, + "0.6": 0.7951807228915663, + "0.7": 0.8469553891240639, + "0.8": 0.8212308694236405, + "0.9": 0.7342885053728427 + }, + "matched_cos_abs": { + "0.5": 0.8948516249656677, + "0.6": 0.8929741382598877, + "0.7": 0.8961443305015564, + "0.8": 0.897276759147644, + "0.9": 0.8950138092041016 + }, + "unique_frac_abs": { + "0.5": 0.7082383588407685, + "0.6": 0.8739824161510908, + "0.7": 0.9586453923803321, + "0.8": 0.9874633669814392, + "0.9": 0.9975577987626181 + }, + "dedup_dcs_abs": { + "0.5": 0.6337682462716794, + "0.6": 0.7804436949168149, + "0.7": 0.8590846333430745, + "0.8": 0.8860279297021265, + "0.9": 0.8928280053717894 + }, + "matched_frac_abs": { + "0.5": 0.6571149462715727, + "0.6": 0.7932269619016606, + "0.7": 0.8472810159557147, + "0.8": 0.8218821230869424, + "0.9": 0.7349397590361446 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.8988218307495117, + "dcs_abs": 0.8989831805229187, + "frac_sign_flipped": 0.006184895833333333, + "alive_source": "stats", + "n_alive_a": 3070, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.9010449051856995, + "0.6": 0.899488091468811, + "0.7": 0.9000774025917053, + "0.8": 0.9009464979171753, + "0.9": 0.8996071815490723 + }, + "unique_frac_signed": { + "0.5": 0.7131227613155324, + "0.6": 0.8785411917942039, + "0.7": 0.9588082057961576, + "0.8": 0.9876261803972647, + "0.9": 0.9980462390100945 + }, + "dedup_dcs_signed": { + "0.5": 0.6425556308553181, + "0.6": 0.7902373398837031, + "0.7": 0.8630015994566187, + "0.8": 0.889798348480232, + "0.9": 0.8978495641315228 + }, + "matched_frac_signed": { + "0.5": 0.6619993487463367, + "0.6": 0.8036470205144904, + "0.7": 0.8489091501139694, + "0.8": 0.832627808531423, + "0.9": 0.7600130250732661 + }, + "matched_cos_abs": { + "0.5": 0.9010125994682312, + "0.6": 0.899365246295929, + "0.7": 0.9005137085914612, + "0.8": 0.9012415409088135, + "0.9": 0.89975905418396 + }, + "unique_frac_abs": { + "0.5": 0.7101921198306741, + "0.6": 0.8749592966460437, + "0.7": 0.9581569521328558, + "0.8": 0.9874633669814392, + "0.9": 0.9980462390100945 + }, + "dedup_dcs_abs": { + "0.5": 0.6398920480104893, + "0.6": 0.7869079833269819, + "0.7": 0.8628334703778491, + "0.8": 0.8899430064493574, + "0.9": 0.898001140043581 + }, + "matched_frac_abs": { + "0.5": 0.6584174535981765, + "0.6": 0.8010420058612829, + "0.7": 0.8485835232823185, + "0.8": 0.8329534353630739, + "0.9": 0.760664278736568 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.9008050560951233, + "dcs_abs": 0.9010428786277771, + "frac_sign_flipped": 0.005208333333333333, + "alive_source": "stats", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.9044643640518188, + "0.6": 0.9022323489189148, + "0.7": 0.9006729125976562, + "0.8": 0.9007866978645325, + "0.9": 0.9012545347213745 + }, + "unique_frac_signed": { + "0.5": 0.712890625, + "0.6": 0.8780924479166666, + "0.7": 0.9606119791666666, + "0.8": 0.9892578125, + "0.9": 0.9978841145833334 + }, + "dedup_dcs_signed": { + "0.5": 0.6447841657791287, + "0.6": 0.792243411851814, + "0.7": 0.8651971891522408, + "0.8": 0.8911102782585658, + "0.9": 0.8993475833946529 + }, + "matched_frac_signed": { + "0.5": 0.6666666666666666, + "0.6": 0.8046875, + "0.7": 0.8541666666666666, + "0.8": 0.8336588541666666, + "0.9": 0.748046875 + }, + "matched_cos_abs": { + "0.5": 0.9049298167228699, + "0.6": 0.9024344086647034, + "0.7": 0.9006250500679016, + "0.8": 0.9010184407234192, + "0.9": 0.9014777541160583 + }, + "unique_frac_abs": { + "0.5": 0.7099609375, + "0.6": 0.8743489583333334, + "0.7": 0.9597981770833334, + "0.8": 0.9892578125, + "0.9": 0.9978841145833334 + }, + "dedup_dcs_abs": { + "0.5": 0.6424648210522719, + "0.6": 0.789042585180141, + "0.7": 0.8644182812907578, + "0.8": 0.8913395316922106, + "0.9": 0.8995703304826748 + }, + "matched_frac_abs": { + "0.5": 0.6647135416666666, + "0.6": 0.8024088541666666, + "0.7": 0.853515625, + "0.8": 0.8346354166666666, + "0.9": 0.7486979166666666 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 1, + "n_dead_std": 1.1547005383792517, + "n_dense_mean": 11, + "n_dense_std": 1.0, + "l0_mean": 31, + "l0_std": 0.27583679987973647, + "recon_r2_mean": 0.8446263020105448, + "recon_r2_std": null, + "l1_per_token_mean": 26.687852151303478, + "l1_per_token_std": 0.07981449373837719, + "babel_mu1_signed_mean": 0.5451075236002604, + "babel_mu1_signed_std": 0.0018654223436076266, + "n_dup_signed_per_threshold": { + "0.5": 1543.0, + "0.6": 679.3333333333334, + "0.7": 230.0, + "0.8": 68.33333333333333, + "0.9": 10.0 + }, + "n_dup_signed_band": 246.91666666666669, + "n_dup_signed_band_std": 5.305264680799001, + "n_dup_clusters_signed_per_threshold": { + "0.5": 658.6666666666666, + "0.6": 304.0, + "0.7": 105.66666666666667, + "0.8": 32.0, + "0.9": 3.3333333333333335 + }, + "n_dup_clusters_signed_band": 111.25, + "n_dup_clusters_signed_band_std": 1.984313483298443, + "largest_dup_signed_per_threshold": { + "0.5": 8.666666666666666, + "0.6": 6.333333333333333, + "0.7": 5.666666666666667, + "0.8": 4.666666666666667, + "0.9": 4.666666666666667 + }, + "largest_dup_signed_band": 5.333333333333334, + "largest_dup_signed_band_std": 0.5204164998665333, + "dcs_signed_mean": 0.8981410264968872, + "dcs_signed_std": 0.0030617365677448848, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.6411633010808654, + "0.6": 0.788386759972624, + "0.7": 0.8626500693499931, + "0.8": 0.8889562995557108, + "0.9": 0.8966422890138791 + }, + "dedup_dcs_signed_band": 0.8591588544730517, + "dedup_dcs_signed_band_std": 0.003471445673971341, + "matched_cos_signed_per_threshold": { + "0.5": 0.900144100189209, + "0.6": 0.8980314135551453, + "0.7": 0.8989430864651998, + "0.8": 0.8995978434880575, + "0.9": 0.8985923329989115 + }, + "matched_cos_signed_band": 0.8987911691268285, + "matched_cos_signed_band_std": 0.0032470658129767666, + "unique_frac_signed_per_threshold": { + "0.5": 0.7122855866031695, + "0.6": 0.877903170254215, + "0.7": 0.9596265481407612, + "0.8": 0.9881700577648432, + "0.9": 0.997829384118682 + }, + "unique_frac_signed_band": 0.9558822900696254, + "unique_frac_signed_band_std": 0.000527445019082348, + "matched_frac_signed_per_threshold": { + "0.5": 0.6629038677231448, + "0.6": 0.8011717478020189, + "0.7": 0.8500104019682332, + "0.8": 0.8291725107072434, + "0.9": 0.7474494684820363 + }, + "matched_frac_signed_band": 0.806951032239883, + "matched_frac_signed_band_std": 0.006553058260641227, + "babel_mu1_abs_mean": 0.5470974246660868, + "babel_mu1_abs_std": 0.0017057661119239966, + "n_dup_abs_per_threshold": { + "0.5": 1558.0, + "0.6": 699.3333333333334, + "0.7": 234.0, + "0.8": 69.0, + "0.9": 10.0 + }, + "n_dup_abs_band": 253.08333333333334, + "n_dup_abs_band_std": 4.481443219916251, + "n_dup_clusters_abs_per_threshold": { + "0.5": 665.0, + "0.6": 313.3333333333333, + "0.7": 107.33333333333333, + "0.8": 32.333333333333336, + "0.9": 3.3333333333333335 + }, + "n_dup_clusters_abs_band": 114.08333333333331, + "n_dup_clusters_abs_band_std": 1.5275252316519468, + "largest_dup_abs_per_threshold": { + "0.5": 8.666666666666666, + "0.6": 6.333333333333333, + "0.7": 5.666666666666667, + "0.8": 4.666666666666667, + "0.9": 4.666666666666667 + }, + "largest_dup_abs_band": 5.333333333333334, + "largest_dup_abs_band_std": 0.5204164998665333, + "dcs_abs_mean": 0.8983086546262106, + "dcs_abs_std": 0.003126542974681518, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.6387083717781469, + "0.6": 0.7854647544746459, + "0.7": 0.8621121283372272, + "0.8": 0.8891034892812315, + "0.9": 0.8967998252993484 + }, + "dedup_dcs_abs_band": 0.8583700493481132, + "dedup_dcs_abs_band_std": 0.0033735010969585706, + "matched_cos_abs_per_threshold": { + "0.5": 0.9002646803855896, + "0.6": 0.8982579310735067, + "0.7": 0.8990943630536398, + "0.8": 0.8998455802599589, + "0.9": 0.8987502058347067 + }, + "matched_cos_abs_band": 0.8989870200554531, + "matched_cos_abs_band_std": 0.003201604256915303, + "unique_frac_abs_per_threshold": { + "0.5": 0.7094638053904809, + "0.6": 0.874430223710156, + "0.7": 0.9588668405321737, + "0.8": 0.9880615154876261, + "0.9": 0.997829384118682 + }, + "unique_frac_abs_band": 0.9547969909621594, + "unique_frac_abs_band_std": 0.0004710052768031956, + "matched_frac_abs_per_threshold": { + "0.5": 0.6600819805121386, + "0.6": 0.79889260730987, + "0.7": 0.8497933880793443, + "0.8": 0.8298236583722276, + "0.9": 0.7481006514797931 + }, + "matched_frac_abs_band": 0.8066525763103088, + "matched_frac_abs_band_std": 0.006358946800346112, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/seed0/config.json b/lm/pythia-160m_layer8/relu/x4/l0_32/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..920481756aae71a20da4a01e54beb062205c045a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/seed0/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..37f4b3c41f29583767abbd975d137bee1451ea16 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 31.55881953125, + "l0_per_token_0.001": 31.503378466796875, + "l0_per_token_0.01": 31.009649658203124, + "l0_per_token_0.05": 28.928172607421875, + "l0_per_token_0.1": 26.56762353515625, + "l0_per_token_0.5": 14.353348291015624, + "l0_per_token_1.0": 7.66328310546875, + "l1_per_token": 26.77954580725691, + "r2_score": 0.8450262789430625, + "l2_loss": 0.10317700737088512 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/seed1/config.json b/lm/pythia-160m_layer8/relu/x4/l0_32/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2b729114fa7a39fd6d4550e7ebd1e771051ed38a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/seed1/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..5d3492d2ee2944099b12f7c2d3f59cfa956d4605 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 31.1541205078125, + "l0_per_token_0.001": 31.099614111328126, + "l0_per_token_0.01": 30.614100048828124, + "l0_per_token_0.05": 28.568532861328126, + "l0_per_token_0.1": 26.2486109375, + "l0_per_token_0.5": 14.238789404296876, + "l0_per_token_1.0": 7.64779052734375, + "l1_per_token": 26.650040147504185, + "r2_score": 0.8443971747611574, + "l2_loss": 0.10359584668358142 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/seed2/config.json b/lm/pythia-160m_layer8/relu/x4/l0_32/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d2ac614f301c676dc8383e32f921621c96254746 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/seed2/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..503dc858299561facdaa7c3a70c72c81673a0fe3 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 31.03178271484375, + "l0_per_token_0.001": 30.97721328125, + "l0_per_token_0.01": 30.4909302734375, + "l0_per_token_0.05": 28.441397607421877, + "l0_per_token_0.1": 26.117450439453126, + "l0_per_token_0.5": 14.100814208984374, + "l0_per_token_1.0": 7.526739501953125, + "l1_per_token": 26.633970499149342, + "r2_score": 0.8444554523274143, + "l2_loss": 0.10355704717072059 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/code_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x4/l0_64/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..d60eea124651b6ee8289f1cd75e078db2d63b477 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 55.478425, + "recon_r2": 0.8714025679466708, + "l1_per_token": 34.09881615998363 + }, + "seed1": { + "l0": 54.66657, + "recon_r2": 0.8702204071291837, + "l1_per_token": 33.757322023278526 + }, + "seed2": { + "l0": 54.49111, + "recon_r2": 0.8700447606479409, + "l1_per_token": 33.6249155148147 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.8146965028975108, + "mcc_abs": 0.8147075916158207, + "act_iou": 0.5293977327988327, + "act_iou_n_pairs": 3072, + "input_r2": 0.984624992395231, + "matched_mcc_signed": { + "0.6": 0.872645704352213, + "0.7": 0.9033813255450119, + "0.8": 0.929609377101118, + "0.9": 0.9555429201224537 + }, + "matched_mcc_abs": { + "0.6": 0.8735128384091018, + "0.7": 0.9014772069806459, + "0.8": 0.9295748640633567, + "0.9": 0.9556145443225736 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.8092405581009112, + "mcc_abs": 0.8092520321548872, + "act_iou": 0.5223443815690956, + "act_iou_n_pairs": 3072, + "input_r2": 0.9843824545215993, + "matched_mcc_signed": { + "0.6": 0.8713554036240858, + "0.7": 0.8996173617898816, + "0.8": 0.9265556471666416, + "0.9": 0.9540283119137328 + }, + "matched_mcc_abs": { + "0.6": 0.8699235354161734, + "0.7": 0.8978648913432846, + "0.8": 0.9264925553655707, + "0.9": 0.9540338917467643 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.813804598992269, + "mcc_abs": 0.8138155142262088, + "act_iou": 0.5305717722597642, + "act_iou_n_pairs": 3072, + "input_r2": 0.9846896218757415, + "matched_mcc_signed": { + "0.6": 0.8748716772096637, + "0.7": 0.9031880128669986, + "0.8": 0.9327310049622786, + "0.9": 0.9567521627190859 + }, + "matched_mcc_abs": { + "0.6": 0.8742229585320231, + "0.7": 0.9031161307138751, + "0.8": 0.9330333142911217, + "0.9": 0.9566831454056545 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 54.878701666666664, + "l0_std": 0.5267329592007078, + "recon_r2_mean": 0.8705559119079318, + "recon_r2_std": 0.000738466490467355, + "l1_per_token_mean": 33.82701789935895, + "l1_per_token_std": 0.2445170581571854, + "input_r2_mean": 0.9845656895975239, + "input_r2_std": 0.00016194308328727454, + "mcc_signed_mean": 0.8125805533302303, + "mcc_signed_std": 0.0029266959601147997, + "mcc_abs_mean": 0.8125917126656389, + "mcc_abs_std": 0.002926439808484948, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8729575950619876, + "0.7": 0.9020622334006307, + "0.8": 0.929632009743346, + "0.9": 0.9554411315850908 + }, + "matched_mcc_signed_band": 0.9150232424477638, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8725531107857661, + "0.7": 0.9008194096792685, + "0.8": 0.9297002445733497, + "0.9": 0.9554438604916641 + }, + "matched_mcc_abs_band": 0.9146291563825121, + "act_iou_mean": 0.5274379622092309, + "act_iou_std": 0.004450057861098715, + "act_iou_n_pairs": 3072.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/dict_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x4/l0_64/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..8c58b4913052682999423e5a274708f02aba8791 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/dict_metrics_step_final.json @@ -0,0 +1,595 @@ +{ + "schema": 3, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 3072, + "d_in": 768, + "l0": 55, + "l0_raw": 55.45199516601563, + "recon_r2": 0.8719332669602384, + "l1_per_token": 34.024666632472126, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 24, + "dense_rate": 0.0078125, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1015, + "0.6": 391, + "0.7": 118, + "0.8": 39, + "0.9": 12 + }, + "n_dup_clusters_signed": { + "0.5": 448, + "0.6": 177, + "0.7": 52, + "0.8": 15, + "0.9": 3 + }, + "n_dup_removed_signed": { + "0.5": 567, + "0.6": 214, + "0.7": 66, + "0.8": 24, + "0.9": 9 + }, + "largest_dup_signed": { + "0.5": 10, + "0.6": 9, + "0.7": 9, + "0.8": 8, + "0.9": 8 + }, + "babel_mu1_signed": 0.49146509170532227, + "n_dup_abs": { + "0.5": 1045, + "0.6": 406, + "0.7": 123, + "0.8": 39, + "0.9": 12 + }, + "n_dup_clusters_abs": { + "0.5": 461, + "0.6": 183, + "0.7": 54, + "0.8": 15, + "0.9": 3 + }, + "n_dup_removed_abs": { + "0.5": 584, + "0.6": 223, + "0.7": 69, + "0.8": 24, + "0.9": 9 + }, + "largest_dup_abs": { + "0.5": 10, + "0.6": 9, + "0.7": 9, + "0.8": 8, + "0.9": 8 + }, + "babel_mu1_abs": 0.4955556392669678, + "dup_excludes_dead": true + }, + "seed1": { + "d_sae": 3072, + "d_in": 768, + "l0": 55, + "l0_raw": 54.660304638671874, + "recon_r2": 0.8709207039956828, + "l1_per_token": 33.687655444244434, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 26, + "dense_rate": 0.008463541666666666, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1007, + "0.6": 357, + "0.7": 132, + "0.8": 34, + "0.9": 13 + }, + "n_dup_clusters_signed": { + "0.5": 439, + "0.6": 163, + "0.7": 60, + "0.8": 14, + "0.9": 4 + }, + "n_dup_removed_signed": { + "0.5": 568, + "0.6": 194, + "0.7": 72, + "0.8": 20, + "0.9": 9 + }, + "largest_dup_signed": { + "0.5": 9, + "0.6": 7, + "0.7": 7, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_signed": 0.4898470342159271, + "n_dup_abs": { + "0.5": 1034, + "0.6": 365, + "0.7": 134, + "0.8": 34, + "0.9": 13 + }, + "n_dup_clusters_abs": { + "0.5": 454, + "0.6": 166, + "0.7": 61, + "0.8": 14, + "0.9": 4 + }, + "n_dup_removed_abs": { + "0.5": 580, + "0.6": 199, + "0.7": 73, + "0.8": 20, + "0.9": 9 + }, + "largest_dup_abs": { + "0.5": 10, + "0.6": 7, + "0.7": 7, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_abs": 0.49384069442749023, + "dup_excludes_dead": true + }, + "seed2": { + "d_sae": 3072, + "d_in": 768, + "l0": 54, + "l0_raw": 54.4679453125, + "recon_r2": 0.9125313644733196, + "l1_per_token": 33.556283884556734, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 24, + "dense_rate": 0.0078125, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 962, + "0.6": 381, + "0.7": 116, + "0.8": 30, + "0.9": 8 + }, + "n_dup_clusters_signed": { + "0.5": 424, + "0.6": 174, + "0.7": 53, + "0.8": 13, + "0.9": 3 + }, + "n_dup_removed_signed": { + "0.5": 538, + "0.6": 207, + "0.7": 63, + "0.8": 17, + "0.9": 5 + }, + "largest_dup_signed": { + "0.5": 6, + "0.6": 5, + "0.7": 5, + "0.8": 4, + "0.9": 4 + }, + "babel_mu1_signed": 0.4880087673664093, + "n_dup_abs": { + "0.5": 1009, + "0.6": 396, + "0.7": 119, + "0.8": 30, + "0.9": 8 + }, + "n_dup_clusters_abs": { + "0.5": 444, + "0.6": 181, + "0.7": 55, + "0.8": 13, + "0.9": 3 + }, + "n_dup_removed_abs": { + "0.5": 565, + "0.6": 215, + "0.7": 64, + "0.8": 17, + "0.9": 5 + }, + "largest_dup_abs": { + "0.5": 6, + "0.6": 5, + "0.7": 5, + "0.8": 4, + "0.9": 4 + }, + "babel_mu1_abs": 0.4922177791595459, + "dup_excludes_dead": true + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.8853657245635986, + "dcs_abs": 0.8855684399604797, + "frac_sign_flipped": 0.0078125, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.8825798630714417, + "0.6": 0.8844672441482544, + "0.7": 0.8859406113624573, + "0.8": 0.8860536217689514, + "0.9": 0.8851085901260376 + }, + "unique_frac_signed": { + "0.5": 0.8152669270833334, + "0.6": 0.93359375, + "0.7": 0.9775390625, + "0.8": 0.9928385416666666, + "0.9": 0.9970703125 + }, + "dedup_dcs_signed": { + "0.5": 0.7195381728718834, + "0.6": 0.8257330912165344, + "0.7": 0.8660415546619333, + "0.8": 0.879708185675554, + "0.9": 0.8825154985534027 + }, + "matched_frac_signed": { + "0.5": 0.763671875, + "0.6": 0.8382161458333334, + "0.7": 0.8434244791666666, + "0.8": 0.8020833333333334, + "0.9": 0.7158203125 + }, + "matched_cos_abs": { + "0.5": 0.8846625089645386, + "0.6": 0.885380744934082, + "0.7": 0.8858501315116882, + "0.8": 0.8862488865852356, + "0.9": 0.8853217363357544 + }, + "unique_frac_abs": { + "0.5": 0.810546875, + "0.6": 0.9313151041666666, + "0.7": 0.9768880208333334, + "0.8": 0.9928385416666666, + "0.9": 0.9970703125 + }, + "dedup_dcs_abs": { + "0.5": 0.7170604320708662, + "0.6": 0.8245684606954455, + "0.7": 0.8653763817274012, + "0.8": 0.8799020521109924, + "0.9": 0.8827280203113332 + }, + "matched_frac_abs": { + "0.5": 0.7568359375, + "0.6": 0.8343098958333334, + "0.7": 0.8440755208333334, + "0.8": 0.8030598958333334, + "0.9": 0.7154947916666666 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.8817655444145203, + "dcs_abs": 0.8820450901985168, + "frac_sign_flipped": 0.007486979166666667, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.8837628364562988, + "0.6": 0.8802902698516846, + "0.7": 0.8819006681442261, + "0.8": 0.8831005096435547, + "0.9": 0.8824936747550964 + }, + "unique_frac_signed": { + "0.5": 0.8201497395833334, + "0.6": 0.9314778645833334, + "0.7": 0.97900390625, + "0.8": 0.9933268229166666, + "0.9": 0.9977213541666666 + }, + "dedup_dcs_signed": { + "0.5": 0.7248178601730615, + "0.6": 0.8199709007749334, + "0.7": 0.8633841990376823, + "0.8": 0.8772074235603213, + "0.9": 0.8804827842201727 + }, + "matched_frac_signed": { + "0.5": 0.7604166666666666, + "0.6": 0.837890625, + "0.7": 0.8434244791666666, + "0.8": 0.7991536458333334, + "0.9": 0.7044270833333334 + }, + "matched_cos_abs": { + "0.5": 0.8839060068130493, + "0.6": 0.880961537361145, + "0.7": 0.8823547959327698, + "0.8": 0.8833776712417603, + "0.9": 0.8827574849128723 + }, + "unique_frac_abs": { + "0.5": 0.81298828125, + "0.6": 0.9287109375, + "0.7": 0.9783528645833334, + "0.8": 0.9933268229166666, + "0.9": 0.9977213541666666 + }, + "dedup_dcs_abs": { + "0.5": 0.7186052252654918, + "0.6": 0.8181586152641103, + "0.7": 0.8632543421796679, + "0.8": 0.8774827356101014, + "0.9": 0.8807459932480317 + }, + "matched_frac_abs": { + "0.5": 0.7542317708333334, + "0.6": 0.8356119791666666, + "0.7": 0.8424479166666666, + "0.8": 0.7998046875, + "0.9": 0.7047526041666666 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.8843539357185364, + "dcs_abs": 0.8845224380493164, + "frac_sign_flipped": 0.005859375, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 3072, + "n_alive_b": 3072, + "matched_cos_signed": { + "0.5": 0.8859319090843201, + "0.6": 0.8826544880867004, + "0.7": 0.8845993876457214, + "0.8": 0.8848937749862671, + "0.9": 0.8849505186080933 + }, + "unique_frac_signed": { + "0.5": 0.8199869791666666, + "0.6": 0.9347330729166666, + "0.7": 0.97802734375, + "0.8": 0.9939778645833334, + "0.9": 0.9977213541666666 + }, + "dedup_dcs_signed": { + "0.5": 0.7264526298774096, + "0.6": 0.8250463419729688, + "0.7": 0.8651623893820215, + "0.8": 0.8795648248439345, + "0.9": 0.8829340297961608 + }, + "matched_frac_signed": { + "0.5": 0.7626953125, + "0.6": 0.8395182291666666, + "0.7": 0.8404947916666666, + "0.8": 0.7942708333333334, + "0.9": 0.7122395833333334 + }, + "matched_cos_abs": { + "0.5": 0.8857351541519165, + "0.6": 0.8835398554801941, + "0.7": 0.884735643863678, + "0.8": 0.885064423084259, + "0.9": 0.8851024508476257 + }, + "unique_frac_abs": { + "0.5": 0.8136393229166666, + "0.6": 0.9326171875, + "0.7": 0.9777018229166666, + "0.8": 0.9939778645833334, + "0.9": 0.9977213541666666 + }, + "dedup_dcs_abs": { + "0.5": 0.7206689511076547, + "0.6": 0.8240044550620951, + "0.7": 0.8650076518048687, + "0.8": 0.8797344452759717, + "0.9": 0.8830856158359286 + }, + "matched_frac_abs": { + "0.5": 0.7581380208333334, + "0.6": 0.8382161458333334, + "0.7": 0.8388671875, + "0.8": 0.7939453125, + "0.9": 0.7125651041666666 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 25, + "n_dense_std": 1.1547005383792515, + "l0_mean": 55, + "l0_std": 0.5215569824501854, + "recon_r2_mean": 0.8851284451430802, + "recon_r2_std": 0.023737024072195766, + "l1_per_token_mean": 33.756201987091096, + "l1_per_token_std": 0.2415979532992064, + "babel_mu1_signed_mean": 0.48977363109588623, + "babel_mu1_signed_std": 0.0017293309392549298, + "n_dup_signed_per_threshold": { + "0.5": 994.6666666666666, + "0.6": 376.3333333333333, + "0.7": 122.0, + "0.8": 34.333333333333336, + "0.9": 11.0 + }, + "n_dup_signed_band": 135.91666666666666, + "n_dup_signed_band_std": 3.538478957593691, + "n_dup_clusters_signed_per_threshold": { + "0.5": 437.0, + "0.6": 171.33333333333334, + "0.7": 55.0, + "0.8": 14.0, + "0.9": 3.3333333333333335 + }, + "n_dup_clusters_signed_band": 60.91666666666667, + "n_dup_clusters_signed_band_std": 0.7637626158259734, + "largest_dup_signed_per_threshold": { + "0.5": 8.333333333333334, + "0.6": 7.0, + "0.7": 7.0, + "0.8": 6.0, + "0.9": 6.0 + }, + "largest_dup_signed_band": 6.5, + "largest_dup_signed_band_std": 2.0, + "dcs_signed_mean": 0.8838284015655518, + "dcs_signed_std": 0.0018567347376293236, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.7236028876407848, + "0.6": 0.8235834446548121, + "0.7": 0.8648627143605457, + "0.8": 0.8788268113599367, + "0.9": 0.8819774375232453 + }, + "dedup_dcs_signed_band": 0.862312601974635, + "dedup_dcs_signed_band_std": 0.0017837681035786535, + "matched_cos_signed_per_threshold": { + "0.5": 0.8840915362040201, + "0.6": 0.8824706673622131, + "0.7": 0.8841468890508016, + "0.8": 0.8846826354662577, + "0.9": 0.8841842611630758 + }, + "matched_cos_signed_band": 0.8838711132605871, + "matched_cos_signed_band_std": 0.001758181543260972, + "unique_frac_signed_per_threshold": { + "0.5": 0.8184678819444445, + "0.6": 0.9332682291666666, + "0.7": 0.9781901041666666, + "0.8": 0.9933810763888888, + "0.9": 0.9975043402777777 + }, + "unique_frac_signed_band": 0.9755859375, + "unique_frac_signed_band_std": 0.00046215074428713164, + "matched_frac_signed_per_threshold": { + "0.5": 0.7622612847222222, + "0.6": 0.8385416666666666, + "0.7": 0.8424479166666666, + "0.8": 0.7985026041666666, + "0.9": 0.7108289930555557 + }, + "matched_frac_signed_band": 0.797580295138889, + "matched_frac_signed_band_std": 0.0020071951750961384, + "babel_mu1_abs_mean": 0.49387137095133465, + "babel_mu1_abs_std": 0.0016691414892141333, + "n_dup_abs_per_threshold": { + "0.5": 1029.3333333333333, + "0.6": 389.0, + "0.7": 125.33333333333333, + "0.8": 34.333333333333336, + "0.9": 11.0 + }, + "n_dup_abs_band": 139.91666666666669, + "n_dup_abs_band_std": 4.488411003165077, + "n_dup_clusters_abs_per_threshold": { + "0.5": 453.0, + "0.6": 176.66666666666666, + "0.7": 56.666666666666664, + "0.8": 14.0, + "0.9": 3.3333333333333335 + }, + "n_dup_clusters_abs_band": 62.666666666666664, + "n_dup_clusters_abs_band_std": 1.282900359861721, + "largest_dup_abs_per_threshold": { + "0.5": 8.666666666666666, + "0.6": 7.0, + "0.7": 7.0, + "0.8": 6.0, + "0.9": 6.0 + }, + "largest_dup_abs_band": 6.5, + "largest_dup_abs_band_std": 2.0, + "dcs_abs_mean": 0.8840453227361044, + "dcs_abs_std": 0.0018094827031107767, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.7187782028146709, + "0.6": 0.8222438436738836, + "0.7": 0.8645461252373127, + "0.8": 0.8790397443323551, + "0.9": 0.8821865431317645 + }, + "dedup_dcs_abs_band": 0.8620040640938289, + "dedup_dcs_abs_band_std": 0.0018155231023493329, + "matched_cos_abs_per_threshold": { + "0.5": 0.8847678899765015, + "0.6": 0.8832940459251404, + "0.7": 0.8843135237693787, + "0.8": 0.884896993637085, + "0.9": 0.8843938906987509 + }, + "matched_cos_abs_band": 0.8842246135075887, + "matched_cos_abs_band_std": 0.0017019007054859593, + "unique_frac_abs_per_threshold": { + "0.5": 0.8123914930555555, + "0.6": 0.9308810763888888, + "0.7": 0.9776475694444445, + "0.8": 0.9933810763888888, + "0.9": 0.9975043402777777 + }, + "unique_frac_abs_band": 0.974853515625, + "unique_frac_abs_band_std": 0.0005638186222554939, + "matched_frac_abs_per_threshold": { + "0.5": 0.7564019097222223, + "0.6": 0.8360460069444445, + "0.7": 0.841796875, + "0.8": 0.7989366319444445, + "0.9": 0.7109375 + }, + "matched_frac_abs_band": 0.7969292534722223, + "matched_frac_abs_band_std": 0.00200058529378173, + "alive_source": "stats", + "frac_denominator": "d_sae", + "d_sae": 3072 + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/seed0/config.json b/lm/pythia-160m_layer8/relu/x4/l0_64/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..920481756aae71a20da4a01e54beb062205c045a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/seed0/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..d768996249d8e9af87530cb8c6814d1ec50a4529 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 55.45199516601563, + "l0_per_token_0.001": 55.31884814453125, + "l0_per_token_0.01": 54.13694541015625, + "l0_per_token_0.05": 49.23156567382812, + "l0_per_token_0.1": 43.82738720703125, + "l0_per_token_0.5": 19.208117578125, + "l0_per_token_1.0": 8.794631494140624, + "l1_per_token": 34.024666632472126, + "r2_score": 0.8719332669602384, + "l2_loss": 0.08526311537653528 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/seed1/config.json b/lm/pythia-160m_layer8/relu/x4/l0_64/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2b729114fa7a39fd6d4550e7ebd1e771051ed38a --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/seed1/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..7bc4464f6c72cb7fb336d6cf3e32817452ff1775 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 54.660304638671874, + "l0_per_token_0.001": 54.529213232421874, + "l0_per_token_0.01": 53.36413129882813, + "l0_per_token_0.05": 48.53085366210937, + "l0_per_token_0.1": 43.206820751953124, + "l0_per_token_0.5": 18.96840849609375, + "l0_per_token_1.0": 8.716828564453126, + "l1_per_token": 33.687655444244434, + "r2_score": 0.8709207039956828, + "l2_loss": 0.0859372504217863 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/seed2/config.json b/lm/pythia-160m_layer8/relu/x4/l0_64/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..d2ac614f301c676dc8383e32f921621c96254746 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/seed2/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 4, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..997efa11db9505fddf82fd7ea2c70fb148b94466 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 54.4679453125, + "l0_per_token_0.001": 54.33675297851563, + "l0_per_token_0.01": 53.17172216796875, + "l0_per_token_0.05": 48.34060048828125, + "l0_per_token_0.1": 43.022510302734375, + "l0_per_token_0.5": 18.84040576171875, + "l0_per_token_1.0": 8.6139404296875, + "l1_per_token": 33.556283884556734, + "r2_score": 0.9125313644733196, + "l2_loss": 0.08611509622179343 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x4/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x8/l0_128/code_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x8/l0_128/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..2f714c84373f0fecb98c8412d1b84c0f2bcb4579 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_128/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 122.887005, + "recon_r2": 0.9097930769937822, + "l1_per_token": 44.564508491802364 + }, + "seed1": { + "l0": 122.95676833333333, + "recon_r2": 0.9095126837716605, + "l1_per_token": 44.6368576795842 + }, + "seed2": { + "l0": 122.17187, + "recon_r2": 0.9094523961307048, + "l1_per_token": 44.32600494884503 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.7354987699813949, + "mcc_abs": 0.7355017451486209, + "act_iou": 0.43363591787864575, + "act_iou_n_pairs": 6144, + "input_r2": 0.9832015740418398, + "matched_mcc_signed": { + "0.6": 0.8494207496884197, + "0.7": 0.8902364115814076, + "0.8": 0.9232312607240409, + "0.9": 0.9543092588516985 + }, + "matched_mcc_abs": { + "0.6": 0.848207747815131, + "0.7": 0.8890888262747284, + "0.8": 0.9233395621365551, + "0.9": 0.9542485366643677 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7349876624216435, + "mcc_abs": 0.734990465267764, + "act_iou": 0.43268764813281263, + "act_iou_n_pairs": 6144, + "input_r2": 0.9833088995988184, + "matched_mcc_signed": { + "0.6": 0.84686694805958, + "0.7": 0.8883211748428332, + "0.8": 0.921114763285595, + "0.9": 0.9540847723508952 + }, + "matched_mcc_abs": { + "0.6": 0.8464274976719964, + "0.7": 0.8863203330865812, + "0.8": 0.9209561778886227, + "0.9": 0.9540826877547313 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7379775376677257, + "mcc_abs": 0.7379786650766457, + "act_iou": 0.4346559945487787, + "act_iou_n_pairs": 6144, + "input_r2": 0.9832662876444279, + "matched_mcc_signed": { + "0.6": 0.8481927839178633, + "0.7": 0.8890118213088146, + "0.8": 0.9233921154625154, + "0.9": 0.9535156654542263 + }, + "matched_mcc_abs": { + "0.6": 0.8472366251453753, + "0.7": 0.8887746211338888, + "0.8": 0.9232195337194145, + "0.9": 0.9535298051862884 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 122.67188111111112, + "l0_std": 0.43442498097527776, + "recon_r2_mean": 0.9095860522987159, + "recon_r2_std": 0.00018180502795161726, + "l1_per_token_mean": 44.50912370674387, + "l1_per_token_std": 0.16265903253118744, + "input_r2_mean": 0.983258920428362, + "input_r2_std": 5.404073185672757e-05, + "mcc_signed_mean": 0.7361546566902547, + "mcc_signed_std": 0.0015992120021835672, + "mcc_abs_mean": 0.7361569584976769, + "mcc_abs_std": 0.0015982218006384307, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8481601605552876, + "0.7": 0.8891898025776852, + "0.8": 0.9225793798240504, + "0.9": 0.9539698988856067 + }, + "matched_mcc_signed_band": 0.9034748104606575, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8472906235441675, + "0.7": 0.8880612601650663, + "0.8": 0.9225050912481975, + "0.9": 0.9539536765351291 + }, + "matched_mcc_abs_band": 0.9029526628731401, + "act_iou_mean": 0.433659853520079, + "act_iou_std": 0.0009843914818352068, + "act_iou_n_pairs": 6144.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x8/l0_128/dict_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x8/l0_128/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..170b35807f2bec90f28623f7b0f178282e2aa595 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_128/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 6144, + "d_in": 768, + "l0": 123, + "l0_raw": 122.7903529296875, + "recon_r2": 0.9101844750520658, + "l1_per_token": 44.45847347420369, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 36, + "dense_rate": 0.005859375, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1333, + "0.6": 480, + "0.7": 167, + "0.8": 37, + "0.9": 11 + }, + "n_dup_clusters_signed": { + "0.5": 591, + "0.6": 218, + "0.7": 78, + "0.8": 15, + "0.9": 3 + }, + "largest_dup_signed": { + "0.5": 7, + "0.6": 7, + "0.7": 6, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_signed": 0.44852808117866516, + "n_dup_abs": { + "0.5": 1421, + "0.6": 510, + "0.7": 185, + "0.8": 39, + "0.9": 11 + }, + "n_dup_clusters_abs": { + "0.5": 632, + "0.6": 231, + "0.7": 87, + "0.8": 16, + "0.9": 3 + }, + "largest_dup_abs": { + "0.5": 8, + "0.6": 7, + "0.7": 6, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_abs": 0.46097955107688904 + }, + "seed1": { + "d_sae": 6144, + "d_in": 768, + "l0": 123, + "l0_raw": 122.8485640625, + "recon_r2": 0.9100155486423547, + "l1_per_token": 44.51770745628815, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 40, + "dense_rate": 0.006510416666666667, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1323, + "0.6": 477, + "0.7": 165, + "0.8": 39, + "0.9": 13 + }, + "n_dup_clusters_signed": { + "0.5": 577, + "0.6": 215, + "0.7": 74, + "0.8": 15, + "0.9": 3 + }, + "largest_dup_signed": { + "0.5": 10, + "0.6": 9, + "0.7": 8, + "0.8": 8, + "0.9": 8 + }, + "babel_mu1_signed": 0.4473077952861786, + "n_dup_abs": { + "0.5": 1421, + "0.6": 516, + "0.7": 186, + "0.8": 45, + "0.9": 13 + }, + "n_dup_clusters_abs": { + "0.5": 623, + "0.6": 234, + "0.7": 84, + "0.8": 18, + "0.9": 3 + }, + "largest_dup_abs": { + "0.5": 13, + "0.6": 9, + "0.7": 8, + "0.8": 8, + "0.9": 8 + }, + "babel_mu1_abs": 0.4605676233768463 + }, + "seed2": { + "d_sae": 6144, + "d_in": 768, + "l0": 122, + "l0_raw": 122.10298837890625, + "recon_r2": 0.9098951397383302, + "l1_per_token": 44.22595241535068, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 36, + "dense_rate": 0.005859375, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 1332, + "0.6": 466, + "0.7": 159, + "0.8": 33, + "0.9": 14 + }, + "n_dup_clusters_signed": { + "0.5": 590, + "0.6": 211, + "0.7": 73, + "0.8": 13, + "0.9": 4 + }, + "largest_dup_signed": { + "0.5": 8, + "0.6": 7, + "0.7": 7, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_signed": 0.4481460750102997, + "n_dup_abs": { + "0.5": 1432, + "0.6": 517, + "0.7": 188, + "0.8": 36, + "0.9": 14 + }, + "n_dup_clusters_abs": { + "0.5": 639, + "0.6": 236, + "0.7": 87, + "0.8": 14, + "0.9": 4 + }, + "largest_dup_abs": { + "0.5": 9, + "0.6": 7, + "0.7": 7, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_abs": 0.461302250623703 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.8267329335212708, + "dcs_abs": 0.8270637392997742, + "frac_sign_flipped": 0.009440104166666666, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.8159302473068237, + "0.6": 0.8221226930618286, + "0.7": 0.8254250288009644, + "0.8": 0.8264380097389221, + "0.9": 0.8267216086387634 + }, + "unique_frac_signed": { + "0.5": 0.87890625, + "0.6": 0.9573567708333334, + "0.7": 0.9853515625, + "0.8": 0.9962565104166666, + "0.9": 0.99853515625 + }, + "dedup_dcs_signed": { + "0.5": 0.717126193922013, + "0.6": 0.7870647266584759, + "0.7": 0.8133338418556377, + "0.8": 0.8233442476581937, + "0.9": 0.825510590657359 + }, + "matched_frac_signed": { + "0.5": 0.7639973958333334, + "0.6": 0.7701822916666666, + "0.7": 0.7366536458333334, + "0.8": 0.6778971354166666, + "0.9": 0.5745442708333334 + }, + "matched_cos_abs": { + "0.5": 0.8183266520500183, + "0.6": 0.8232998251914978, + "0.7": 0.8261932730674744, + "0.8": 0.8268958926200867, + "0.9": 0.8270229697227478 + }, + "unique_frac_abs": { + "0.5": 0.870849609375, + "0.6": 0.954345703125, + "0.7": 0.9837239583333334, + "0.8": 0.9959309895833334, + "0.9": 0.99853515625 + }, + "dedup_dcs_abs": { + "0.5": 0.71263944527891, + "0.6": 0.7857126505550696, + "0.7": 0.8127461169303085, + "0.8": 0.8235312446195167, + "0.9": 0.825811510294443 + }, + "matched_frac_abs": { + "0.5": 0.759765625, + "0.6": 0.77001953125, + "0.7": 0.7361653645833334, + "0.8": 0.6778971354166666, + "0.9": 0.5748697916666666 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.8263147473335266, + "dcs_abs": 0.8265665173530579, + "frac_sign_flipped": 0.0078125, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.8153768181800842, + "0.6": 0.8213448524475098, + "0.7": 0.8244780898094177, + "0.8": 0.8259482383728027, + "0.9": 0.8261820673942566 + }, + "unique_frac_signed": { + "0.5": 0.8792317708333334, + "0.6": 0.9579264322916666, + "0.7": 0.9857584635416666, + "0.8": 0.99658203125, + "0.9": 0.99853515625 + }, + "dedup_dcs_signed": { + "0.5": 0.7169052037449244, + "0.6": 0.7867879441861684, + "0.7": 0.8127362550342998, + "0.8": 0.8231251731049269, + "0.9": 0.824971839756472 + }, + "matched_frac_signed": { + "0.5": 0.7654622395833334, + "0.6": 0.7716471354166666, + "0.7": 0.7366536458333334, + "0.8": 0.67724609375, + "0.9": 0.5703125 + }, + "matched_cos_abs": { + "0.5": 0.8176923990249634, + "0.6": 0.8220580816268921, + "0.7": 0.825183093547821, + "0.8": 0.8261095881462097, + "0.9": 0.8264146447181702 + }, + "unique_frac_abs": { + "0.5": 0.8712565104166666, + "0.6": 0.9544270833333334, + "0.7": 0.9838053385416666, + "0.8": 0.996337890625, + "0.9": 0.99853515625 + }, + "dedup_dcs_abs": { + "0.5": 0.7124198261687221, + "0.6": 0.7845944971777499, + "0.7": 0.8118195327066738, + "0.8": 0.8230842844786821, + "0.9": 0.8252040763909463 + }, + "matched_frac_abs": { + "0.5": 0.7586263020833334, + "0.6": 0.7703450520833334, + "0.7": 0.7374674479166666, + "0.8": 0.6769205729166666, + "0.9": 0.5703125 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.8285920023918152, + "dcs_abs": 0.8288652300834656, + "frac_sign_flipped": 0.007975260416666666, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.8181618452072144, + "0.6": 0.8244134783744812, + "0.7": 0.8272290825843811, + "0.8": 0.8285210728645325, + "0.9": 0.8282991051673889 + }, + "unique_frac_signed": { + "0.5": 0.87890625, + "0.6": 0.9579264322916666, + "0.7": 0.985595703125, + "0.8": 0.9964192708333334, + "0.9": 0.9983723958333334 + }, + "dedup_dcs_signed": { + "0.5": 0.7190875592641532, + "0.6": 0.7897274620724298, + "0.7": 0.8153134292952018, + "0.8": 0.8255543632937286, + "0.9": 0.8269509620925722 + }, + "matched_frac_signed": { + "0.5": 0.7669270833333334, + "0.6": 0.77685546875, + "0.7": 0.74072265625, + "0.8": 0.6787109375, + "0.9": 0.5799153645833334 + }, + "matched_cos_abs": { + "0.5": 0.8201780319213867, + "0.6": 0.8251090049743652, + "0.7": 0.8276336193084717, + "0.8": 0.8289664387702942, + "0.9": 0.8285712599754333 + }, + "unique_frac_abs": { + "0.5": 0.8705240885416666, + "0.6": 0.9541829427083334, + "0.7": 0.9834798177083334, + "0.8": 0.9960123697916666, + "0.9": 0.9983723958333334 + }, + "dedup_dcs_abs": { + "0.5": 0.7139847336802632, + "0.6": 0.7873049384215847, + "0.7": 0.813960961046784, + "0.8": 0.8256608271573592, + "0.9": 0.8272226739403171 + }, + "matched_frac_abs": { + "0.5": 0.76123046875, + "0.6": 0.7752278645833334, + "0.7": 0.73974609375, + "0.8": 0.67822265625, + "0.9": 0.5797526041666666 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 37, + "n_dense_std": 2.309401076758503, + "l0_mean": 123, + "l0_std": 0.4146769147220374, + "recon_r2_mean": 0.9100317211442502, + "recon_r2_std": null, + "l1_per_token_mean": 44.40071111528084, + "l1_per_token_std": 0.15421614298837652, + "babel_mu1_signed_mean": 0.4479939838250478, + "babel_mu1_signed_std": 0.0006241980545659582, + "n_dup_signed_per_threshold": { + "0.5": 1329.3333333333333, + "0.6": 474.3333333333333, + "0.7": 163.66666666666666, + "0.8": 36.333333333333336, + "0.9": 12.666666666666666 + }, + "n_dup_signed_band": 171.75, + "n_dup_signed_band_std": 3.25, + "n_dup_clusters_signed_per_threshold": { + "0.5": 586.0, + "0.6": 214.66666666666666, + "0.7": 75.0, + "0.8": 14.333333333333334, + "0.9": 3.3333333333333335 + }, + "n_dup_clusters_signed_band": 76.83333333333331, + "n_dup_clusters_signed_band_std": 1.626601774661928, + "largest_dup_signed_per_threshold": { + "0.5": 8.333333333333334, + "0.6": 7.666666666666667, + "0.7": 7.0, + "0.8": 7.0, + "0.9": 7.0 + }, + "largest_dup_signed_band": 7.166666666666667, + "largest_dup_signed_band_std": 1.0103629710818451, + "dcs_signed_mean": 0.8272132277488708, + "dcs_signed_std": 0.0012122229823447108, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.7177063189770303, + "0.6": 0.7878600443056913, + "0.7": 0.8137945087283797, + "0.8": 0.8240079280189496, + "0.9": 0.8258111308354678 + }, + "dedup_dcs_signed_band": 0.812868402972122, + "dedup_dcs_signed_band_std": 0.0013304936187497896, + "matched_cos_signed_per_threshold": { + "0.5": 0.8164896368980408, + "0.6": 0.8226270079612732, + "0.7": 0.8257107337315878, + "0.8": 0.8269691069920858, + "0.9": 0.8270675937334696 + }, + "matched_cos_signed_band": 0.8255936106046041, + "matched_cos_signed_band_std": 0.0013623686255255052, + "unique_frac_signed_per_threshold": { + "0.5": 0.8790147569444445, + "0.6": 0.9577365451388888, + "0.7": 0.9855685763888888, + "0.8": 0.9964192708333334, + "0.9": 0.9984809027777778 + }, + "unique_frac_signed_band": 0.9845513237847222, + "unique_frac_signed_band_std": 0.00016444709815782187, + "matched_frac_signed_per_threshold": { + "0.5": 0.7654622395833334, + "0.6": 0.7728949652777777, + "0.7": 0.738009982638889, + "0.8": 0.6779513888888888, + "0.9": 0.574924045138889 + }, + "matched_frac_signed_band": 0.690945095486111, + "matched_frac_signed_band_std": 0.002723604009440633, + "babel_mu1_abs_mean": 0.4609498083591461, + "babel_mu1_abs_std": 0.0003682156567829661, + "n_dup_abs_per_threshold": { + "0.5": 1424.6666666666667, + "0.6": 514.3333333333334, + "0.7": 186.33333333333334, + "0.8": 40.0, + "0.9": 12.666666666666666 + }, + "n_dup_abs_band": 188.33333333333334, + "n_dup_abs_band_std": 1.9094065395649336, + "n_dup_clusters_abs_per_threshold": { + "0.5": 631.3333333333334, + "0.6": 233.66666666666666, + "0.7": 86.0, + "0.8": 16.0, + "0.9": 3.3333333333333335 + }, + "n_dup_clusters_abs_band": 84.74999999999999, + "n_dup_clusters_abs_band_std": 0.5, + "largest_dup_abs_per_threshold": { + "0.5": 10.0, + "0.6": 7.666666666666667, + "0.7": 7.0, + "0.8": 7.0, + "0.9": 7.0 + }, + "largest_dup_abs_band": 7.166666666666667, + "largest_dup_abs_band_std": 1.0103629710818451, + "dcs_abs_mean": 0.8274984955787659, + "dcs_abs_std": 0.0012094543483846978, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.713014668375965, + "0.6": 0.7858706953848014, + "0.7": 0.8128422035612554, + "0.8": 0.8240921187518526, + "0.9": 0.8260794202085688 + }, + "dedup_dcs_abs_band": 0.8122211094766195, + "dedup_dcs_abs_band_std": 0.001203926599031214, + "matched_cos_abs_per_threshold": { + "0.5": 0.8187323609987894, + "0.6": 0.8234889705975851, + "0.7": 0.826336661974589, + "0.8": 0.8273239731788635, + "0.9": 0.8273362914721171 + }, + "matched_cos_abs_band": 0.8261214743057886, + "matched_cos_abs_band_std": 0.001334772082814074, + "unique_frac_abs_per_threshold": { + "0.5": 0.870876736111111, + "0.6": 0.954318576388889, + "0.7": 0.9836697048611112, + "0.8": 0.99609375, + "0.9": 0.9984809027777778 + }, + "unique_frac_abs_band": 0.9831407335069444, + "unique_frac_abs_band_std": 0.0001323731912973195, + "matched_frac_abs_per_threshold": { + "0.5": 0.7598741319444445, + "0.6": 0.7718641493055557, + "0.7": 0.73779296875, + "0.8": 0.6776801215277777, + "0.9": 0.574978298611111 + }, + "matched_frac_abs_band": 0.6905788845486112, + "matched_frac_abs_band_std": 0.00235346910012415, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x8/l0_32/code_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x8/l0_32/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..0d9cfdf95dfc04510d10ce783987ef34a799bda2 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_32/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 34.74431333333333, + "recon_r2": 0.8591056644109554, + "l1_per_token": 26.980237984298245 + }, + "seed1": { + "l0": 34.2245, + "recon_r2": 0.8578357936400712, + "l1_per_token": 26.75897234018537 + }, + "seed2": { + "l0": 34.32180833333334, + "recon_r2": 0.8580524394832058, + "l1_per_token": 26.721103489418947 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.7961923176432123, + "mcc_abs": 0.7961992181360054, + "act_iou": 0.4929559235236222, + "act_iou_n_pairs": 6144, + "input_r2": 0.9829441483814376, + "matched_mcc_signed": { + "0.6": 0.8420141889220589, + "0.7": 0.8794588281235379, + "0.8": 0.9108134763102883, + "0.9": 0.9455462141853819 + }, + "matched_mcc_abs": { + "0.6": 0.8414154700748735, + "0.7": 0.8795114763348252, + "0.8": 0.9108515462553312, + "0.9": 0.9455462125873666 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7973017955437712, + "mcc_abs": 0.7973099982889597, + "act_iou": 0.4940935105170592, + "act_iou_n_pairs": 6144, + "input_r2": 0.9834155010699716, + "matched_mcc_signed": { + "0.6": 0.8451226280033737, + "0.7": 0.8786327358180824, + "0.8": 0.9084980557946704, + "0.9": 0.9450638338630089 + }, + "matched_mcc_abs": { + "0.6": 0.8451147776237812, + "0.7": 0.8773086428936129, + "0.8": 0.9086079034815396, + "0.9": 0.9450008461767898 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7935804677785957, + "mcc_abs": 0.7935845417571246, + "act_iou": 0.4914903331355256, + "act_iou_n_pairs": 6144, + "input_r2": 0.9829637362510482, + "matched_mcc_signed": { + "0.6": 0.841462198204437, + "0.7": 0.879415021619617, + "0.8": 0.9100669718266049, + "0.9": 0.9454806724608373 + }, + "matched_mcc_abs": { + "0.6": 0.8419668479240305, + "0.7": 0.879834125721526, + "0.8": 0.9102730871959611, + "0.9": 0.945373724495061 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 34.43020722222222, + "l0_std": 0.2763407583257237, + "recon_r2_mean": 0.8583312991780775, + "recon_r2_std": 0.0006793121453017185, + "l1_per_token_mean": 26.820104604634185, + "l1_per_token_std": 0.13996619922884015, + "input_r2_mean": 0.9831077952341524, + "input_r2_std": 0.0002666609874106489, + "mcc_signed_mean": 0.7956915269885264, + "mcc_signed_std": 0.0019105401183519071, + "mcc_abs_mean": 0.7956979193940299, + "mcc_abs_std": 0.0019126504945174518, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8428663383766232, + "0.7": 0.8791688618537458, + "0.8": 0.9097928346438545, + "0.9": 0.945363573503076 + }, + "matched_mcc_signed_band": 0.8942979020943249, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8428323652075619, + "0.7": 0.8788847483166546, + "0.8": 0.9099108456442773, + "0.9": 0.9453069277530725 + }, + "matched_mcc_abs_band": 0.8942337217303916, + "act_iou_mean": 0.4928465890587357, + "act_iou_std": 0.0013050282138103557, + "act_iou_n_pairs": 6144.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x8/l0_32/dict_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x8/l0_32/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..413574cff105765689a5ea6384bc7687d15a6046 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_32/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 6144, + "d_in": 768, + "l0": 35, + "l0_raw": 34.74092470703125, + "recon_r2": 0.8597877585491807, + "l1_per_token": 26.927988527501665, + "stats_source": "stats", + "n_dead": 2, + "dead_rate": 0.0003255208333333333, + "alive_source": "stats", + "n_dense": 12, + "dense_rate": 0.001953125, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 3665, + "0.6": 1788, + "0.7": 589, + "0.8": 191, + "0.9": 16 + }, + "n_dup_clusters_signed": { + "0.5": 1528, + "0.6": 798, + "0.7": 264, + "0.8": 89, + "0.9": 6 + }, + "largest_dup_signed": { + "0.5": 12, + "0.6": 7, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_signed": 0.5718369483947754, + "n_dup_abs": { + "0.5": 3683, + "0.6": 1804, + "0.7": 595, + "0.8": 191, + "0.9": 16 + }, + "n_dup_clusters_abs": { + "0.5": 1540, + "0.6": 805, + "0.7": 267, + "0.8": 89, + "0.9": 6 + }, + "largest_dup_abs": { + "0.5": 12, + "0.6": 7, + "0.7": 6, + "0.8": 5, + "0.9": 5 + }, + "babel_mu1_abs": 0.5728697776794434 + }, + "seed1": { + "d_sae": 6144, + "d_in": 768, + "l0": 34, + "l0_raw": 34.2394486328125, + "recon_r2": 0.8587411297408798, + "l1_per_token": 26.720831672987465, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 8, + "dense_rate": 0.0013020833333333333, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 3616, + "0.6": 1729, + "0.7": 573, + "0.8": 168, + "0.9": 21 + }, + "n_dup_clusters_signed": { + "0.5": 1505, + "0.6": 772, + "0.7": 265, + "0.8": 79, + "0.9": 8 + }, + "largest_dup_signed": { + "0.5": 9, + "0.6": 6, + "0.7": 6, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_signed": 0.5691758990287781, + "n_dup_abs": { + "0.5": 3638, + "0.6": 1754, + "0.7": 579, + "0.8": 168, + "0.9": 21 + }, + "n_dup_clusters_abs": { + "0.5": 1511, + "0.6": 783, + "0.7": 268, + "0.8": 79, + "0.9": 8 + }, + "largest_dup_abs": { + "0.5": 9, + "0.6": 6, + "0.7": 6, + "0.8": 6, + "0.9": 6 + }, + "babel_mu1_abs": 0.5703765749931335 + }, + "seed2": { + "d_sae": 6144, + "d_in": 768, + "l0": 34, + "l0_raw": 34.3545751953125, + "recon_r2": 0.8587780545584602, + "l1_per_token": 26.68443019515859, + "stats_source": "stats", + "n_dead": 1, + "dead_rate": 0.00016276041666666666, + "alive_source": "stats", + "n_dense": 10, + "dense_rate": 0.0016276041666666667, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 3654, + "0.6": 1829, + "0.7": 588, + "0.8": 165, + "0.9": 20 + }, + "n_dup_clusters_signed": { + "0.5": 1519, + "0.6": 811, + "0.7": 273, + "0.8": 79, + "0.9": 9 + }, + "largest_dup_signed": { + "0.5": 10, + "0.6": 8, + "0.7": 4, + "0.8": 4, + "0.9": 3 + }, + "babel_mu1_signed": 0.5722100138664246, + "n_dup_abs": { + "0.5": 3668, + "0.6": 1854, + "0.7": 594, + "0.8": 165, + "0.9": 20 + }, + "n_dup_clusters_abs": { + "0.5": 1526, + "0.6": 823, + "0.7": 275, + "0.8": 79, + "0.9": 9 + }, + "largest_dup_abs": { + "0.5": 10, + "0.6": 8, + "0.7": 4, + "0.8": 4, + "0.9": 3 + }, + "babel_mu1_abs": 0.573321521282196 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.8903469443321228, + "dcs_abs": 0.8905156254768372, + "frac_sign_flipped": 0.00390625, + "alive_source": "stats", + "n_alive_a": 6142, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.8932364583015442, + "0.6": 0.8891727328300476, + "0.7": 0.8899311423301697, + "0.8": 0.8906798958778381, + "0.9": 0.8903899192810059 + }, + "unique_frac_signed": { + "0.5": 0.6542405990558359, + "0.6": 0.8415269412339248, + "0.7": 0.9484779423734332, + "0.8": 0.9844538499104672, + "0.9": 0.9981279505127788 + }, + "dedup_dcs_signed": { + "0.5": 0.5843915555777154, + "0.6": 0.7482628100870797, + "0.7": 0.8440800587313583, + "0.8": 0.8768332525347918, + "0.9": 0.888723065289189 + }, + "matched_frac_signed": { + "0.5": 0.6088230506267296, + "0.6": 0.7611916001953443, + "0.7": 0.825980791144392, + "0.8": 0.8049812795051278, + "0.9": 0.7131694611753215 + }, + "matched_cos_abs": { + "0.5": 0.8927739858627319, + "0.6": 0.8891279697418213, + "0.7": 0.8900185823440552, + "0.8": 0.8908162713050842, + "0.9": 0.8905608057975769 + }, + "unique_frac_abs": { + "0.5": 0.6524499430245808, + "0.6": 0.8396548917467036, + "0.7": 0.9479895816376364, + "0.8": 0.9844538499104672, + "0.9": 0.9981279505127788 + }, + "dedup_dcs_abs": { + "0.5": 0.5824903362099674, + "0.6": 0.7465606491825353, + "0.7": 0.8437283435260631, + "0.8": 0.8769675078491774, + "0.9": 0.8888936318977443 + }, + "matched_frac_abs": { + "0.5": 0.6075207553312714, + "0.6": 0.7600520918118183, + "0.7": 0.8258180042324597, + "0.8": 0.8049812795051278, + "0.9": 0.7138206088230507 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.8918004035949707, + "dcs_abs": 0.8919652104377747, + "frac_sign_flipped": 0.00439453125, + "alive_source": "stats", + "n_alive_a": 6142, + "n_alive_b": 6143, + "matched_cos_signed": { + "0.5": 0.8930925130844116, + "0.6": 0.8911546468734741, + "0.7": 0.8912075757980347, + "0.8": 0.8922520279884338, + "0.9": 0.8918845653533936 + }, + "unique_frac_signed": { + "0.5": 0.6523402523402524, + "0.6": 0.8366300366300367, + "0.7": 0.947985347985348, + "0.8": 0.9846967846967847, + "0.9": 0.9982905982905983 + }, + "dedup_dcs_signed": { + "0.5": 0.5826001953486752, + "0.6": 0.745566744856782, + "0.7": 0.8448517238700782, + "0.8": 0.8785977030993963, + "0.9": 0.8903599763527895 + }, + "matched_frac_signed": { + "0.5": 0.6098494098494098, + "0.6": 0.7581603581603582, + "0.7": 0.8297924297924298, + "0.8": 0.8074888074888075, + "0.9": 0.7156695156695156 + }, + "matched_cos_abs": { + "0.5": 0.8932340741157532, + "0.6": 0.8916364908218384, + "0.7": 0.8912026286125183, + "0.8": 0.8923702239990234, + "0.9": 0.8920436501502991 + }, + "unique_frac_abs": { + "0.5": 0.6512820512820513, + "0.6": 0.8348392348392348, + "0.7": 0.9474155474155475, + "0.8": 0.9846967846967847, + "0.9": 0.9982905982905983 + }, + "dedup_dcs_abs": { + "0.5": 0.5817473200651315, + "0.6": 0.744373125752444, + "0.7": 0.8443392262451038, + "0.8": 0.8787140903309879, + "0.9": 0.8905187892098713 + }, + "matched_frac_abs": { + "0.5": 0.6072446072446073, + "0.6": 0.7565323565323565, + "0.7": 0.8301180301180301, + "0.8": 0.807000407000407, + "0.9": 0.715995115995116 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.889320433139801, + "dcs_abs": 0.8894422054290771, + "frac_sign_flipped": 0.0030924479166666665, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6143, + "matched_cos_signed": { + "0.5": 0.894135594367981, + "0.6": 0.8905164003372192, + "0.7": 0.8888174891471863, + "0.8": 0.8888114094734192, + "0.9": 0.8894379734992981 + }, + "unique_frac_signed": { + "0.5": 0.6545128998128102, + "0.6": 0.839342394400586, + "0.7": 0.9493773907381786, + "0.8": 0.985757304468137, + "0.9": 0.9980467160413445 + }, + "dedup_dcs_signed": { + "0.5": 0.5852232806956379, + "0.6": 0.7474481677120324, + "0.7": 0.843823228689015, + "0.8": 0.8761523391830432, + "0.9": 0.8877006485734429 + }, + "matched_frac_signed": { + "0.5": 0.6074713111418573, + "0.6": 0.7591763652641003, + "0.7": 0.8254252461951657, + "0.8": 0.8008464230487508, + "0.9": 0.7116464556034834 + }, + "matched_cos_abs": { + "0.5": 0.8940474987030029, + "0.6": 0.8907026648521423, + "0.7": 0.8889794945716858, + "0.8": 0.8889538645744324, + "0.9": 0.889557957649231 + }, + "unique_frac_abs": { + "0.5": 0.6526410026857654, + "0.6": 0.8371449499470985, + "0.7": 0.9488076829169041, + "0.8": 0.985757304468137, + "0.9": 0.9980467160413445 + }, + "dedup_dcs_abs": { + "0.5": 0.5834920560022284, + "0.6": 0.745647237785394, + "0.7": 0.8434705744052017, + "0.8": 0.8762927653394258, + "0.9": 0.8878203983602604 + }, + "matched_frac_abs": { + "0.5": 0.6064946691625295, + "0.6": 0.7568975339790022, + "0.7": 0.8244486042158379, + "0.8": 0.801009196711972, + "0.9": 0.7121347765931473 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 1, + "n_dead_std": 1.0, + "n_dense_mean": 10, + "n_dense_std": 2.0, + "l0_mean": 34, + "l0_std": 0.26267797890580524, + "recon_r2_mean": 0.8591023142828402, + "recon_r2_std": null, + "l1_per_token_mean": 26.777750131882573, + "l1_per_token_std": 0.13137712116828038, + "babel_mu1_signed_mean": 0.5710742870966593, + "babel_mu1_signed_std": 0.0016546003758429646, + "n_dup_signed_per_threshold": { + "0.5": 3645.0, + "0.6": 1782.0, + "0.7": 583.3333333333334, + "0.8": 174.66666666666666, + "0.9": 19.0 + }, + "n_dup_signed_band": 639.75, + "n_dup_signed_band_std": 14.89337100860648, + "n_dup_clusters_signed_per_threshold": { + "0.5": 1517.3333333333333, + "0.6": 793.6666666666666, + "0.7": 267.3333333333333, + "0.8": 82.33333333333333, + "0.9": 7.666666666666667 + }, + "n_dup_clusters_signed_band": 287.75, + "n_dup_clusters_signed_band_std": 6.139014578904337, + "largest_dup_signed_per_threshold": { + "0.5": 10.333333333333334, + "0.6": 7.0, + "0.7": 5.333333333333333, + "0.8": 5.0, + "0.9": 4.666666666666667 + }, + "largest_dup_signed_band": 5.5, + "largest_dup_signed_band_std": 0.6614378277661477, + "dcs_signed_mean": 0.8904892603556315, + "dcs_signed_std": 0.001246095402663505, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.5840716772073429, + "0.6": 0.7470925742186315, + "0.7": 0.8442516704301505, + "0.8": 0.8771944316057438, + "0.9": 0.8889278967384738 + }, + "dedup_dcs_signed_band": 0.8393666432482499, + "dedup_dcs_signed_band_std": 0.0005396607876078595, + "matched_cos_signed_per_threshold": { + "0.5": 0.8934881885846456, + "0.6": 0.8902812600135803, + "0.7": 0.8899854024251302, + "0.8": 0.8905811111132304, + "0.9": 0.8905708193778992 + }, + "matched_cos_signed_band": 0.89035464823246, + "matched_cos_signed_band_std": 0.001146572776485083, + "unique_frac_signed_per_threshold": { + "0.5": 0.6536979170696328, + "0.6": 0.8391664574215157, + "0.7": 0.9486135603656534, + "0.8": 0.9849693130251297, + "0.9": 0.9981550882815738 + }, + "unique_frac_signed_band": 0.9427261047734682, + "unique_frac_signed_band_std": 0.0007148717257872185, + "matched_frac_signed_per_threshold": { + "0.5": 0.6087145905393322, + "0.6": 0.7595094412066009, + "0.7": 0.8270661557106624, + "0.8": 0.8044388366808953, + "0.9": 0.7134951441494403 + }, + "matched_frac_signed_band": 0.7761273944368998, + "matched_frac_signed_band_std": 0.0017609091930292323, + "babel_mu1_abs_mean": 0.5721892913182577, + "babel_mu1_abs_std": 0.001586024409391474, + "n_dup_abs_per_threshold": { + "0.5": 3663.0, + "0.6": 1804.0, + "0.7": 589.3333333333334, + "0.8": 174.66666666666666, + "0.9": 19.0 + }, + "n_dup_abs_band": 646.75, + "n_dup_abs_band_std": 14.471955638406303, + "n_dup_clusters_abs_per_threshold": { + "0.5": 1525.6666666666667, + "0.6": 803.6666666666666, + "0.7": 270.0, + "0.8": 82.33333333333333, + "0.9": 7.666666666666667 + }, + "n_dup_clusters_abs_band": 290.91666666666663, + "n_dup_clusters_abs_band_std": 6.043246919771964, + "largest_dup_abs_per_threshold": { + "0.5": 10.333333333333334, + "0.6": 7.0, + "0.7": 5.333333333333333, + "0.8": 5.0, + "0.9": 4.666666666666667 + }, + "largest_dup_abs_band": 5.5, + "largest_dup_abs_band_std": 0.6614378277661477, + "dcs_abs_mean": 0.8906410137812296, + "dcs_abs_std": 0.001266167539718579, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.5825765707591092, + "0.6": 0.7455270042401244, + "0.7": 0.8438460480587895, + "0.8": 0.8773247878398638, + "0.9": 0.889077606489292 + }, + "dedup_dcs_abs_band": 0.8389438616570174, + "dedup_dcs_abs_band_std": 0.0005948394573821579, + "matched_cos_abs_per_threshold": { + "0.5": 0.8933518528938293, + "0.6": 0.8904890418052673, + "0.7": 0.890066901842753, + "0.8": 0.8907134532928467, + "0.9": 0.890720804532369 + }, + "matched_cos_abs_band": 0.890497550368309, + "matched_cos_abs_band_std": 0.0011760513916992082, + "unique_frac_abs_per_threshold": { + "0.5": 0.6521243323307991, + "0.6": 0.8372130255110123, + "0.7": 0.9480709373233628, + "0.8": 0.9849693130251297, + "0.9": 0.9981550882815738 + }, + "unique_frac_abs_band": 0.9421020910352697, + "unique_frac_abs_band_std": 0.000688011057262157, + "matched_frac_abs_per_threshold": { + "0.5": 0.607086677246136, + "0.6": 0.757827327441059, + "0.7": 0.8267948795221093, + "0.8": 0.8043302944058356, + "0.9": 0.7139835004704379 + }, + "matched_frac_abs_band": 0.7757340004598604, + "matched_frac_abs_band_std": 0.001931398147657487, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x8/l0_64/code_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x8/l0_64/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..d17ab84cb82b99fa9528cd6e50edcc1833c71653 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_64/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 61.35798833333333, + "recon_r2": 0.883875598108084, + "l1_per_token": 33.476568084850314 + }, + "seed1": { + "l0": 60.826295, + "recon_r2": 0.8828848448495958, + "l1_per_token": 33.27132824938047 + }, + "seed2": { + "l0": 60.860081666666666, + "recon_r2": 0.8833481666575375, + "l1_per_token": 33.284818533143 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.774333614405394, + "mcc_abs": 0.7743360883399779, + "act_iou": 0.4701510443475343, + "act_iou_n_pairs": 6144, + "input_r2": 0.9831539955621419, + "matched_mcc_signed": { + "0.6": 0.834825383792399, + "0.7": 0.8775079586874162, + "0.8": 0.9152494031821111, + "0.9": 0.9485610019394735 + }, + "matched_mcc_abs": { + "0.6": 0.833599436203003, + "0.7": 0.8776215588674022, + "0.8": 0.9146939135660905, + "0.9": 0.9486339715368957 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7791112216327143, + "mcc_abs": 0.7791158480281529, + "act_iou": 0.47428313766850305, + "act_iou_n_pairs": 6144, + "input_r2": 0.9837094829342937, + "matched_mcc_signed": { + "0.6": 0.8417881831080707, + "0.7": 0.8778438727397928, + "0.8": 0.9137621854752424, + "0.9": 0.9482545346334406 + }, + "matched_mcc_abs": { + "0.6": 0.8414985478644975, + "0.7": 0.8773559367805117, + "0.8": 0.9139690530346546, + "0.9": 0.9483120507991241 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7749974169594601, + "mcc_abs": 0.7749995629320771, + "act_iou": 0.4711445854753333, + "act_iou_n_pairs": 6144, + "input_r2": 0.983181460709816, + "matched_mcc_signed": { + "0.6": 0.8383723153064977, + "0.7": 0.8775884033616874, + "0.8": 0.9127437264050655, + "0.9": 0.9497160258591337 + }, + "matched_mcc_abs": { + "0.6": 0.8373077694905721, + "0.7": 0.8771256733556164, + "0.8": 0.9126447466183525, + "0.9": 0.9497219871503586 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 61.014788333333335, + "l0_std": 0.2976996216173447, + "recon_r2_mean": 0.8833695365384058, + "recon_r2_std": 0.0004957222091630394, + "l1_per_token_mean": 33.3442382891246, + "l1_per_token_std": 0.11479929471301797, + "input_r2_mean": 0.9833483130687505, + "input_r2_std": 0.0003130835949335955, + "mcc_signed_mean": 0.7761474176658562, + "mcc_signed_std": 0.0025880994809259505, + "mcc_abs_mean": 0.776150499766736, + "mcc_abs_std": 0.002589404828760215, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8383286274023224, + "0.7": 0.8776467449296321, + "0.8": 0.9139184383541398, + "0.9": 0.948843854144016 + }, + "matched_mcc_signed_band": 0.8946844162075276, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8374685845193576, + "0.7": 0.8773677230011768, + "0.8": 0.9137692377396993, + "0.9": 0.9488893364954594 + }, + "matched_mcc_abs_band": 0.8943737204389233, + "act_iou_mean": 0.4718595891637902, + "act_iou_std": 0.0021568429495860174, + "act_iou_n_pairs": 6144.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x8/l0_64/dict_metrics_step_final.json b/lm/pythia-160m_layer8/relu/x8/l0_64/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..98985cd73674af2dccc25ae4191479bf285bd6fe --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_64/dict_metrics_step_final.json @@ -0,0 +1,595 @@ +{ + "schema": 3, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 6144, + "d_in": 768, + "l0": 61, + "l0_raw": 61.33215146484375, + "recon_r2": 0.8844004791704476, + "l1_per_token": 33.40253760390999, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 19, + "dense_rate": 0.0030924479166666665, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 2656, + "0.6": 1055, + "0.7": 345, + "0.8": 93, + "0.9": 12 + }, + "n_dup_clusters_signed": { + "0.5": 1152, + "0.6": 470, + "0.7": 156, + "0.8": 41, + "0.9": 3 + }, + "n_dup_removed_signed": { + "0.5": 1504, + "0.6": 585, + "0.7": 189, + "0.8": 52, + "0.9": 9 + }, + "largest_dup_signed": { + "0.5": 8, + "0.6": 8, + "0.7": 8, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_signed": 0.5230711102485657, + "n_dup_abs": { + "0.5": 2716, + "0.6": 1072, + "0.7": 347, + "0.8": 93, + "0.9": 12 + }, + "n_dup_clusters_abs": { + "0.5": 1181, + "0.6": 479, + "0.7": 157, + "0.8": 41, + "0.9": 3 + }, + "n_dup_removed_abs": { + "0.5": 1535, + "0.6": 593, + "0.7": 190, + "0.8": 52, + "0.9": 9 + }, + "largest_dup_abs": { + "0.5": 14, + "0.6": 8, + "0.7": 8, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_abs": 0.5260302424430847, + "dup_excludes_dead": true + }, + "seed1": { + "d_sae": 6144, + "d_in": 768, + "l0": 61, + "l0_raw": 60.824160986328124, + "recon_r2": 0.8835481243786724, + "l1_per_token": 33.20346450786915, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 19, + "dense_rate": 0.0030924479166666665, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 2605, + "0.6": 1031, + "0.7": 353, + "0.8": 90, + "0.9": 17 + }, + "n_dup_clusters_signed": { + "0.5": 1121, + "0.6": 459, + "0.7": 160, + "0.8": 40, + "0.9": 5 + }, + "n_dup_removed_signed": { + "0.5": 1484, + "0.6": 572, + "0.7": 193, + "0.8": 50, + "0.9": 12 + }, + "largest_dup_signed": { + "0.5": 11, + "0.6": 8, + "0.7": 8, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_signed": 0.5221807360649109, + "n_dup_abs": { + "0.5": 2655, + "0.6": 1037, + "0.7": 354, + "0.8": 91, + "0.9": 17 + }, + "n_dup_clusters_abs": { + "0.5": 1146, + "0.6": 462, + "0.7": 160, + "0.8": 40, + "0.9": 5 + }, + "n_dup_removed_abs": { + "0.5": 1509, + "0.6": 575, + "0.7": 194, + "0.8": 51, + "0.9": 12 + }, + "largest_dup_abs": { + "0.5": 11, + "0.6": 8, + "0.7": 8, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_abs": 0.525067150592804, + "dup_excludes_dead": true + }, + "seed2": { + "d_sae": 6144, + "d_in": 768, + "l0": 61, + "l0_raw": 60.85968725585938, + "recon_r2": 0.9215295617691274, + "l1_per_token": 33.21581973728131, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 20, + "dense_rate": 0.0032552083333333335, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 2604, + "0.6": 1031, + "0.7": 361, + "0.8": 100, + "0.9": 14 + }, + "n_dup_clusters_signed": { + "0.5": 1126, + "0.6": 458, + "0.7": 163, + "0.8": 44, + "0.9": 4 + }, + "n_dup_removed_signed": { + "0.5": 1478, + "0.6": 573, + "0.7": 198, + "0.8": 56, + "0.9": 10 + }, + "largest_dup_signed": { + "0.5": 10, + "0.6": 8, + "0.7": 8, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_signed": 0.522265613079071, + "n_dup_abs": { + "0.5": 2661, + "0.6": 1046, + "0.7": 366, + "0.8": 100, + "0.9": 14 + }, + "n_dup_clusters_abs": { + "0.5": 1153, + "0.6": 467, + "0.7": 165, + "0.8": 44, + "0.9": 4 + }, + "n_dup_removed_abs": { + "0.5": 1508, + "0.6": 579, + "0.7": 201, + "0.8": 56, + "0.9": 10 + }, + "largest_dup_abs": { + "0.5": 10, + "0.6": 8, + "0.7": 8, + "0.8": 7, + "0.9": 7 + }, + "babel_mu1_abs": 0.5251511931419373, + "dup_excludes_dead": true + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.8699539303779602, + "dcs_abs": 0.8701098561286926, + "frac_sign_flipped": 0.0048828125, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.86618572473526, + "0.6": 0.8658153414726257, + "0.7": 0.8678197264671326, + "0.8": 0.8692722916603088, + "0.9": 0.8698644638061523 + }, + "unique_frac_signed": { + "0.5": 0.7568359375, + "0.6": 0.9058430989583334, + "0.7": 0.9689127604166666, + "0.8": 0.99169921875, + "0.9": 0.998291015625 + }, + "dedup_dcs_signed": { + "0.5": 0.6555604850291274, + "0.6": 0.7842928520452309, + "0.7": 0.840841606715306, + "0.8": 0.8620566525205504, + "0.9": 0.8683778790291399 + }, + "matched_frac_signed": { + "0.5": 0.6988932291666666, + "0.6": 0.8030598958333334, + "0.7": 0.8046875, + "0.8": 0.7517903645833334, + "0.9": 0.65771484375 + }, + "matched_cos_abs": { + "0.5": 0.8679826855659485, + "0.6": 0.8663020730018616, + "0.7": 0.8679859638214111, + "0.8": 0.8693387508392334, + "0.9": 0.8700197339057922 + }, + "unique_frac_abs": { + "0.5": 0.7522786458333334, + "0.6": 0.9049479166666666, + "0.7": 0.96875, + "0.8": 0.9916178385416666, + "0.9": 0.998291015625 + }, + "dedup_dcs_abs": { + "0.5": 0.6529648393043317, + "0.6": 0.7839582561670492, + "0.7": 0.840861402451992, + "0.8": 0.8620518130677131, + "0.9": 0.8685328837746056 + }, + "matched_frac_abs": { + "0.5": 0.6954752604166666, + "0.6": 0.8037109375, + "0.7": 0.80419921875, + "0.8": 0.75244140625, + "0.9": 0.6575520833333334 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.872567892074585, + "dcs_abs": 0.8728378415107727, + "frac_sign_flipped": 0.005859375, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.8680152893066406, + "0.6": 0.868851900100708, + "0.7": 0.8705528974533081, + "0.8": 0.8720329403877258, + "0.9": 0.8724732398986816 + }, + "unique_frac_signed": { + "0.5": 0.75732421875, + "0.6": 0.90576171875, + "0.7": 0.968505859375, + "0.8": 0.9912109375, + "0.9": 0.9984537760416666 + }, + "dedup_dcs_signed": { + "0.5": 0.6573690008372068, + "0.6": 0.7869727903744206, + "0.7": 0.8431355820794124, + "0.8": 0.8643685883725993, + "0.9": 0.8711242008721456 + }, + "matched_frac_signed": { + "0.5": 0.69873046875, + "0.6": 0.80224609375, + "0.7": 0.8102213541666666, + "0.8": 0.76416015625, + "0.9": 0.6638997395833334 + }, + "matched_cos_abs": { + "0.5": 0.8696610331535339, + "0.6": 0.8696385025978088, + "0.7": 0.870876669883728, + "0.8": 0.8722889423370361, + "0.9": 0.87274169921875 + }, + "unique_frac_abs": { + "0.5": 0.7523600260416666, + "0.6": 0.9046223958333334, + "0.7": 0.9681803385416666, + "0.8": 0.9912109375, + "0.9": 0.9984537760416666 + }, + "dedup_dcs_abs": { + "0.5": 0.6542981975508155, + "0.6": 0.7866944657289423, + "0.7": 0.843165669076067, + "0.8": 0.864622340304777, + "0.9": 0.8713922450939814 + }, + "matched_frac_abs": { + "0.5": 0.6943359375, + "0.6": 0.8019205729166666, + "0.7": 0.8103841145833334, + "0.8": 0.7635091145833334, + "0.9": 0.6640625 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.8699417114257812, + "dcs_abs": 0.870171070098877, + "frac_sign_flipped": 0.00439453125, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.8632369041442871, + "0.6": 0.8645705580711365, + "0.7": 0.8684114813804626, + "0.8": 0.8695156574249268, + "0.9": 0.8698076009750366 + }, + "unique_frac_signed": { + "0.5": 0.7589518229166666, + "0.6": 0.9068196614583334, + "0.7": 0.9681803385416666, + "0.8": 0.9913736979166666, + "0.9": 0.9982096354166666 + }, + "dedup_dcs_signed": { + "0.5": 0.6551552220092465, + "0.6": 0.7840095807769103, + "0.7": 0.8407789220364066, + "0.8": 0.8620149526977912, + "0.9": 0.8682503282519367 + }, + "matched_frac_signed": { + "0.5": 0.6980794270833334, + "0.6": 0.79833984375, + "0.7": 0.806640625, + "0.8": 0.7569986979166666, + "0.9": 0.6529947916666666 + }, + "matched_cos_abs": { + "0.5": 0.86407470703125, + "0.6": 0.8650348782539368, + "0.7": 0.8687326312065125, + "0.8": 0.8696484565734863, + "0.9": 0.8700370192527771 + }, + "unique_frac_abs": { + "0.5": 0.7544759114583334, + "0.6": 0.9060872395833334, + "0.7": 0.9678548177083334, + "0.8": 0.9912923177083334, + "0.9": 0.9982096354166666 + }, + "dedup_dcs_abs": { + "0.5": 0.6519235521554947, + "0.6": 0.7837970649804145, + "0.7": 0.8408070624136599, + "0.8": 0.8620758341082062, + "0.9": 0.868479335787318 + }, + "matched_frac_abs": { + "0.5": 0.6946614583333334, + "0.6": 0.7986653645833334, + "0.7": 0.8068033854166666, + "0.8": 0.75732421875, + "0.9": 0.65283203125 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 19, + "n_dense_std": 0.5773502691896258, + "l0_mean": 61, + "l0_std": 0.28358974795190944, + "recon_r2_mean": 0.8964927217727491, + "recon_r2_std": 0.021686727390870894, + "l1_per_token_mean": 33.27394061635348, + "l1_per_token_std": 0.11153946310572464, + "babel_mu1_signed_mean": 0.5225058197975159, + "babel_mu1_signed_std": 0.0004913918979991983, + "n_dup_signed_per_threshold": { + "0.5": 2621.6666666666665, + "0.6": 1039.0, + "0.7": 353.0, + "0.8": 94.33333333333333, + "0.9": 14.333333333333334 + }, + "n_dup_signed_band": 375.16666666666663, + "n_dup_signed_band_std": 2.0966242709015206, + "n_dup_clusters_signed_per_threshold": { + "0.5": 1133.0, + "0.6": 462.3333333333333, + "0.7": 159.66666666666666, + "0.8": 41.666666666666664, + "0.9": 4.0 + }, + "n_dup_clusters_signed_band": 166.91666666666666, + "n_dup_clusters_signed_band_std": 0.8036375634160795, + "largest_dup_signed_per_threshold": { + "0.5": 9.666666666666666, + "0.6": 8.0, + "0.7": 8.0, + "0.8": 7.0, + "0.9": 7.0 + }, + "largest_dup_signed_band": 7.5, + "largest_dup_signed_band_std": 0.0, + "dcs_signed_mean": 0.8708211779594421, + "dcs_signed_std": 0.0015127111342643934, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.6560282359585269, + "0.6": 0.7850917410655206, + "0.7": 0.8415853702770416, + "0.8": 0.862813397863647, + "0.9": 0.8692508027177407 + }, + "dedup_dcs_signed_band": 0.8396853279809875, + "dedup_dcs_signed_band_std": 0.0014865966510038493, + "matched_cos_signed_per_threshold": { + "0.5": 0.8658126393953959, + "0.6": 0.8664125998814901, + "0.7": 0.8689280351003011, + "0.8": 0.8702736298243204, + "0.9": 0.8707151015599569 + }, + "matched_cos_signed_band": 0.8690823415915172, + "matched_cos_signed_band_std": 0.0016425025862677745, + "unique_frac_signed_per_threshold": { + "0.5": 0.7577039930555555, + "0.6": 0.9061414930555557, + "0.7": 0.968532986111111, + "0.8": 0.9914279513888888, + "0.9": 0.998318142361111 + }, + "unique_frac_signed_band": 0.9661051432291666, + "unique_frac_signed_band_std": 0.0001076558964463201, + "matched_frac_signed_per_threshold": { + "0.5": 0.6985677083333334, + "0.6": 0.8012152777777778, + "0.7": 0.8071831597222222, + "0.8": 0.7576497395833334, + "0.9": 0.658203125 + }, + "matched_frac_signed_band": 0.7560628255208334, + "matched_frac_signed_band_std": 0.0035353589453525134, + "babel_mu1_abs_mean": 0.5254161953926086, + "babel_mu1_abs_std": 0.000533438021390471, + "n_dup_abs_per_threshold": { + "0.5": 2677.3333333333335, + "0.6": 1051.6666666666667, + "0.7": 355.6666666666667, + "0.8": 94.66666666666667, + "0.9": 14.333333333333334 + }, + "n_dup_abs_band": 379.08333333333337, + "n_dup_abs_band_std": 3.7610946987989196, + "n_dup_clusters_abs_per_threshold": { + "0.5": 1160.0, + "0.6": 469.3333333333333, + "0.7": 160.66666666666666, + "0.8": 41.666666666666664, + "0.9": 4.0 + }, + "n_dup_clusters_abs_band": 168.91666666666666, + "n_dup_clusters_abs_band_std": 1.876388374866284, + "largest_dup_abs_per_threshold": { + "0.5": 11.666666666666666, + "0.6": 8.0, + "0.7": 8.0, + "0.8": 7.0, + "0.9": 7.0 + }, + "largest_dup_abs_band": 7.5, + "largest_dup_abs_band_std": 0.0, + "dcs_abs_mean": 0.8710395892461141, + "dcs_abs_std": 0.001557632881345536, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.6530621963368807, + "0.6": 0.7848165956254686, + "0.7": 0.841611377980573, + "0.8": 0.8629166624935655, + "0.9": 0.8694681548853017 + }, + "dedup_dcs_abs_band": 0.8397031977462273, + "dedup_dcs_abs_band_std": 0.0015292593508790364, + "matched_cos_abs_per_threshold": { + "0.5": 0.8672394752502441, + "0.6": 0.8669918179512024, + "0.7": 0.8691984216372172, + "0.8": 0.8704253832499186, + "0.9": 0.8709328174591064 + }, + "matched_cos_abs_band": 0.8693871100743611, + "matched_cos_abs_band_std": 0.0017316512012481072, + "unique_frac_abs_per_threshold": { + "0.5": 0.7530381944444445, + "0.6": 0.9052191840277778, + "0.7": 0.96826171875, + "0.8": 0.9913736979166666, + "0.9": 0.998318142361111 + }, + "unique_frac_abs_band": 0.9657931857638888, + "unique_frac_abs_band_std": 0.00015405024809569753, + "matched_frac_abs_per_threshold": { + "0.5": 0.69482421875, + "0.6": 0.8014322916666666, + "0.7": 0.80712890625, + "0.8": 0.7577582465277778, + "0.9": 0.6581488715277778 + }, + "matched_frac_abs_band": 0.7561170789930555, + "matched_frac_abs_band_std": 0.003348064587457467, + "alive_source": "stats", + "frac_denominator": "d_sae", + "d_sae": 6144 + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x8/l0_64/seed0/config.json b/lm/pythia-160m_layer8/relu/x8/l0_64/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..03af2764551fb0c668e41544170a64c51da927c1 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_64/seed0/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x8/l0_64/seed1/config.json b/lm/pythia-160m_layer8/relu/x8/l0_64/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2df92766c911a9cf15a36365f5f0d91bde86a2e6 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_64/seed1/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..a4968b606c70b1ef0415ebbc1abaac0309eeb515 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 60.824160986328124, + "l0_per_token_0.001": 60.655221533203125, + "l0_per_token_0.01": 59.15923818359375, + "l0_per_token_0.05": 53.01557358398438, + "l0_per_token_0.1": 46.3752421875, + "l0_per_token_0.5": 18.401097998046875, + "l0_per_token_1.0": 8.07302900390625, + "l1_per_token": 33.20346450786915, + "r2_score": 0.8835481243786724, + "l2_loss": 0.07753028027842696 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/relu/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/relu/x8/l0_64/seed2/config.json b/lm/pythia-160m_layer8/relu/x8/l0_64/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..6ff526fe86b1fee16310adf1b05c81222284af8c --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_64/seed2/config.json @@ -0,0 +1,10 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "relu", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "class_type": "VanillaSAE", + "module_name": "fastsae.models.sae" +} diff --git a/lm/pythia-160m_layer8/relu/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/relu/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..30890ad7a4d8bbb807313251d3ff1a8e6cd991c6 --- /dev/null +++ b/lm/pythia-160m_layer8/relu/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 60.85968725585938, + "l0_per_token_0.001": 60.691197119140625, + "l0_per_token_0.01": 59.19782314453125, + "l0_per_token_0.05": 53.063976806640625, + "l0_per_token_0.1": 46.434332275390624, + "l0_per_token_0.5": 18.469414599609376, + "l0_per_token_1.0": 8.08255498046875, + "l1_per_token": 33.21581973728131, + "r2_score": 0.9215295617691274, + "l2_loss": 0.07725614213744848 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/model.safetensors b/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b1a1b96f146832f239b6005edc553e6f6a763a31 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9db9eb958dca739f7c3039399ee021202d295be54979a65cadd6130f20eb1990 +size 75550008 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..f0d5fb7205ea20bf3e6782f9be490385585c40dc --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b1437d53e609ee29a9afe5b8205e636ad92d3a6a94953721dc5fc54807a9f6ba +size 50471 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..28b50fa442efb25d03f43304b2a5cce10634481f --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4e67456e1f6709bc49b45a3e548bde578d2367857295a1c32195c026d826cec +size 50382 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3fb463dcaaa625e454adebc359d750c55733ee91 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5bb3c16d4e597900a3f4c99ad6547bf8bee3c03523500a8df76351291a7d6cfa +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/model.safetensors b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e3b70f71f49e129241785bbdfe4eabb3d075aa1a --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4bd70adbcbb7dd6495c6822f03a6653181d3de9f467194dc6e2b062dbe88bc05 +size 75550008 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..15fc72308a47ed71de3917671ae6d86e0ecfa0d5 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1d0e143d9ffbaf873485e1df60ad7dbfd0d8c3ca3214bbd4fe8924032938705 +size 99648 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..6809b3239f44f098023ccaed3f0a5ade44264945 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17d38cb092da1d4849a235b0bef805d6321c2fb236ed26978d6352199a626f20 +size 50471 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..98087d277e12f5a34282538792f15a64f12cc197 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:734b4d75ea5533ce1bd3600d4c4995bfa56216ffce99250b1d930d8829334f37 +size 50382 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..751ebdc94e4ca77277c958631d499fa94c451d99 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ec24f4673e791a6d599b1dbb3700575c87bd1c904f774fb4a49dcc2b5412c98 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..1754bd2748132cf906270582cacc79b61621ea05 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2e7b0cb643611635eb5b793d17c8acbf7621a0934aa3c2bff82250b52e6e5bf +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/model.safetensors b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0e8491848e27759c850d187e76af46eeba92c78a --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38fd2f6159426268fa32f5c6d94d1be92099c0feae7ded5b6020473e328eb2e2 +size 75550008 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..15fc72308a47ed71de3917671ae6d86e0ecfa0d5 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1d0e143d9ffbaf873485e1df60ad7dbfd0d8c3ca3214bbd4fe8924032938705 +size 99648 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..548b45edf422d2a3f8dd388351c47f38f9df272f --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88f316fc794ebdac8419f0e9aa58ceb8a62f24066045e841c589a2bc78abadf8 +size 50471 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..07c6e2715fb1bd9f46923ebb107c56777100c3eb --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6d856ec257592b80293cc15141c942ea26ba592835f70d3f66c7d13a79d770b +size 50382 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0cc4b947961a89fc8f1c3cac6dc04e75953af1d0 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:469b6327e56174831be8280cf1fb882007d7ef826bfab596aec92c3fd81bfe9a +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..234ecb073cf14588b696bf96566867bd50fd2e70 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_128/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3cda16220a9e7d28fcbb8dae2e7e52301a5a7d3fe49626d02607cca6d329cf95 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/model.safetensors b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..adabc481baa5ad1b70b35184d4ba6ccea05fc9b8 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7023cb8e032efe9e7d6a7cc8b2e953c5890da63f2856713b5a7c485c071e8cab +size 75550008 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..afbab17d2a7c87272f0ca9958f131e0b739dc5fb --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:35aececace33b219d2bd406cb84ddcd74a2bd62111eeca19e185f0abe4b21f70 +size 99648 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..29309c8d64d6535ff453c57cc94cfa42046982ff --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e1d0212671f250ac18dc6626c7ecd850ad86e52cd725795dc6c489a224ef874 +size 50471 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8aa09f8932d66191bdad718ac5b9a48fdc520f84 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:240683463d13b4f63d07ec32bee4837841e054c0ca67d0426c5d00d6dd040c84 +size 50382 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d1f9fcb5c365d3a00081dfaf94dbeb9d1f2891b7 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf3e3552a69fc09c6b46d1423bd9712727d1ecee066adf7e11862c72a83800d5 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0c58ebe990b18ff9be7db448813450d631e6a737 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:abd854c0680cdd511b0d3fc9ff9c92878ac23bfa11bd75e7a89cf4bb6009b638 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/model.safetensors b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..46baefc54d91be3cad550a58b72ca16c3b64c65e --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a6c45032a780131254661b4532006e766d32f2c000aaa4227e8f60e2168940cc +size 75550008 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..afbab17d2a7c87272f0ca9958f131e0b739dc5fb --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:35aececace33b219d2bd406cb84ddcd74a2bd62111eeca19e185f0abe4b21f70 +size 99648 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..679b785b29563544fc12a978f5dd4ec0ffd758a2 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db715e782a2e4e2490ff75587657abf032605e327f4b22cb2162efa8d0fc8315 +size 50471 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..984e808a898700259d436f70a5dda1c007bdd6ff --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5769035ae3483ab3232396de00cc34cb7fc666c43e10808250b294d1601c0e4d +size 50382 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..35862e5645e71fc74c6086194f5d0077a0cc687a --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4c4f06a877b454801d1eea5350af5c40e1f261411e5b2f19789012a2b5caed7 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b0b4f12a82f705845fd3650d0c8700a96859cc96 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd68215c949f6be683c85415250fccb7bde908e9eddc089d6403aa64e6de6b47 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/model.safetensors b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..488e462624a4e226243e278bb6b2b9c397fdfd05 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9d856acbaa1108570e82acf00cd9c05debc89927503076f9adae502df356202 +size 75550008 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..afbab17d2a7c87272f0ca9958f131e0b739dc5fb --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:35aececace33b219d2bd406cb84ddcd74a2bd62111eeca19e185f0abe4b21f70 +size 99648 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..7c955169632585c78f0e781bc324e388faece6eb --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d99b7e799b89713b0d99c9b8097194fb3502ca10bf95c6d6a66dc0711477c1c4 +size 50471 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..072437f97cea6d7b8c9780914db43690218a084d --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58f800ac479c2f27e41f64d3570dac1f908c70cb7b6b7fb8c58387f745d60918 +size 50382 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..630639f888de79357eef21cd84df425e88cbd93a --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d468ca51cdb289e37fc0fc42c8b5186d51201ae44fd866231eb2a7b3ceac080 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..504d12207e29e901cc14c6bc980d7adfd503d0cc --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_32/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe1b6381dc78bbd724cb2dbfd810aca23815994f2fc1a02ebce89d427937f5ff +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/model.safetensors b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4929737dab778e6c426bb1284a248079488a2fe4 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d14fb2aefea95e43b84ac20708b66fb4eda4bd06cdc7ee2d8f504afe9ceb175 +size 75550008 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8de83be649f2f0cc84109b3da72b06a86da913fc --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53a972b5048d9cd93f2f67cbe0716fdc3f9f7eef130ca0bcd46ceccf0b967cde +size 99648 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..3a2edb745e96bbbe63fead3712901e601174cf87 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:845915a008acb87c7e0bcf606ee45e9988440286e6ff26adfe0d3a1f813236e3 +size 50471 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3c2ce3a9b6926fa532c0661486d5eda672a170eb --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efd71b4f782870927c3ae9e076d4abe6518405326552bf20dfceef613744f93c +size 50382 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c4d0fd0a60ed0ae2da451a03e23ce78870294465 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:880a270d9df3d8f891d5f3445ff68f44ad5ceddfcc8fae11ce4d017cc7a32f17 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..440cb4ee6f2467dda0ab1e0d92c1ac942daceb40 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed0/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ea7efd4eaf14f646b324d4c0d155fd0272c8836e49a48082e766ede14a64487 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/model.safetensors b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..51235d740274a97b1b172717bd330bf3c892b612 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:34adb6e9731b2c8a9cec7aa2e2277fcb442e374d539e755160cefd0c1fa50f66 +size 75550008 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8de83be649f2f0cc84109b3da72b06a86da913fc --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53a972b5048d9cd93f2f67cbe0716fdc3f9f7eef130ca0bcd46ceccf0b967cde +size 99648 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..14ebb3aa6dbbe86274dff835c1f65da5662ff72b --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80696bc8f948e7cfa4137462d6107289d2259b66011bd552fb409823956c043f +size 50471 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..be9fc133a67ffb665501b1ed3d49312daf3a2261 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fbaf2e78c5e5902506e2c7f9616bdcf8c640641363cb4cb01f5f3343b122b823 +size 50382 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8019a5276c76c9ce5bf206cd3168bfbae91ca846 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4e987d26d6a03b66b7417505be1a7c13268a5323ba7c231cd38cbdce5e19ad6 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..9119b2b496a0f5e8754b8768367b3fa3831fe65a --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed1/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0efa9cd90a40a2f8256b8ddc9d570fad67428c90f2502ef9b1ceaefd8c6dad2e +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/model.safetensors b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..01b7c23740d8248e06faed7321ed373c7c7bbd6a --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc1e3ee33d2c4979e33b21fad9e5a66fb5baabcd836c7218eea44d742c3fd854 +size 75550008 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8de83be649f2f0cc84109b3da72b06a86da913fc --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53a972b5048d9cd93f2f67cbe0716fdc3f9f7eef130ca0bcd46ceccf0b967cde +size 99648 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..4d1ce04d7a08359ff2730c4daa70b97a163e76fa --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:439260438dd3cbf2b17972cbf1e03cf4a8b2630c8eaf6ff89f291eb41b52adc8 +size 50471 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a9b02e9f2b62b3fb9a2f4251da00a42ba807f185 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27e71768f7af73afabcdb906a2a3511e83c1a98bbfd01d1b2cd032ea5177a9f3 +size 50382 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a8e7b0f42adbe3965cdbd7455ce52f23b47cfa32 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98128b31999e43d29f0baf0a425d2e5f6e3c3fe2b6c83e395ec23559f0c43579 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..27cc569333838246f46da39635f5e7d66a29013f --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x16/l0_64/seed2/stats.fineweb.split_train.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:057e32010477e064a02645a5f0449a8b4c13fc2a52ce039e9537112a8d671918 +size 50387 diff --git a/lm/pythia-160m_layer8/topk/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..077ed98aabac4ec00926dac1e7617f1dd0bf1541 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 128.0, + "l0_per_token_0.05": 128.0, + "l0_per_token_0.1": 127.9995642578125, + "l0_per_token_0.5": 83.25368481445312, + "l0_per_token_1.0": 26.380702392578126, + "l1_per_token": 114.87411509271087, + "r2_score": 0.943075983804663, + "l2_loss": 0.03789835849917313 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x4/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..19089a2223c4b7515b4d50879b5e4d9b5f5642ac --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 128.0, + "l0_per_token_0.05": 128.0, + "l0_per_token_0.1": 127.9995390625, + "l0_per_token_0.5": 83.1790263671875, + "l0_per_token_1.0": 26.465366455078126, + "l1_per_token": 114.70705069301026, + "r2_score": 0.9614652294519404, + "l2_loss": 0.03793846163998607 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..642c940f0af51c624c56af1eb48cf5cc4c88531d --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x4/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 3072, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/code_metrics_step_final.json b/lm/pythia-160m_layer8/topk/x8/l0_128/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..a0073591287fdcda86060d5be04185497439af96 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 128.0, + "recon_r2": 0.9485025733977754, + "l1_per_token": 111.33850701369171 + }, + "seed1": { + "l0": 128.0, + "recon_r2": 0.9481637461910407, + "l1_per_token": 111.48713326041393 + }, + "seed2": { + "l0": 128.0, + "recon_r2": 0.9485215479830029, + "l1_per_token": 111.3216467335262 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.6321417772497461, + "mcc_abs": 0.6321532118100589, + "act_iou": 0.32301991789518647, + "act_iou_n_pairs": 6144, + "input_r2": 0.965758026096733, + "matched_mcc_signed": { + "0.6": 0.8373168194335892, + "0.7": 0.8676057483756484, + "0.8": 0.8975921009527141, + "0.9": 0.9335475883848616 + }, + "matched_mcc_abs": { + "0.6": 0.8390372779282507, + "0.7": 0.8696068019800669, + "0.8": 0.8992253251723297, + "0.9": 0.9348060750603536 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.6330650916355107, + "mcc_abs": 0.6330792819331984, + "act_iou": 0.3241674652688491, + "act_iou_n_pairs": 6144, + "input_r2": 0.9662007245991415, + "matched_mcc_signed": { + "0.6": 0.8388916971229912, + "0.7": 0.8702164182470973, + "0.8": 0.9000037598372457, + "0.9": 0.9333417002317558 + }, + "matched_mcc_abs": { + "0.6": 0.8394208045758106, + "0.7": 0.8712796247229617, + "0.8": 0.9013097755553338, + "0.9": 0.9349110999305237 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.6359908598854371, + "mcc_abs": 0.6360105731128699, + "act_iou": 0.32636757641004277, + "act_iou_n_pairs": 6144, + "input_r2": 0.9659365011586579, + "matched_mcc_signed": { + "0.6": 0.8406483072587141, + "0.7": 0.8721043751072559, + "0.8": 0.9006894203449433, + "0.9": 0.9351677694609604 + }, + "matched_mcc_abs": { + "0.6": 0.8414924471900073, + "0.7": 0.8733933228784644, + "0.8": 0.9015762371499595, + "0.9": 0.9365813533676002 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 128.0, + "l0_std": 0.0, + "recon_r2_mean": 0.948395955857273, + "recon_r2_std": 0.00020132313712022763, + "l1_per_token_mean": 111.38242900254393, + "l1_per_token_std": 0.09106757641385248, + "input_r2_mean": 0.9659650839515108, + "input_r2_std": 0.00022272903502718454, + "mcc_signed_mean": 0.6337325762568979, + "mcc_signed_std": 0.0020094802994645084, + "mcc_abs_mean": 0.6337476889520425, + "mcc_abs_std": 0.0020136745050369715, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8389522746050981, + "0.7": 0.8699755139100006, + "0.8": 0.8994284270449677, + "0.9": 0.9340190193591926 + }, + "matched_mcc_signed_band": 0.8855938087298147, + "matched_mcc_abs_per_threshold": { + "0.6": 0.839983509898023, + "0.7": 0.871426583193831, + "0.8": 0.9007037792925411, + "0.9": 0.9354328427861592 + }, + "matched_mcc_abs_band": 0.8868866787926386, + "act_iou_mean": 0.3245183198580261, + "act_iou_std": 0.0017011844668002353, + "act_iou_n_pairs": 6144.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/dict_metrics_step_final.json b/lm/pythia-160m_layer8/topk/x8/l0_128/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..a3ae1b47159f293baf5e12d4964d84c15afaa39b --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/dict_metrics_step_final.json @@ -0,0 +1,595 @@ +{ + "schema": 3, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 6144, + "d_in": 768, + "l0": 128, + "l0_raw": 128.0, + "recon_r2": 0.9487603959321035, + "l1_per_token": 111.12371754331225, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 308, + "dense_rate": 0.050130208333333336, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 61, + "0.6": 9, + "0.7": 3, + "0.8": 2, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 30, + "0.6": 4, + "0.7": 1, + "0.8": 1, + "0.9": 0 + }, + "n_dup_removed_signed": { + "0.5": 31, + "0.6": 5, + "0.7": 2, + "0.8": 1, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 3, + "0.6": 3, + "0.7": 3, + "0.8": 2, + "0.9": 1 + }, + "babel_mu1_signed": 0.2331564575433731, + "n_dup_abs": { + "0.5": 706, + "0.6": 613, + "0.7": 565, + "0.8": 512, + "0.9": 424 + }, + "n_dup_clusters_abs": { + "0.5": 352, + "0.6": 306, + "0.7": 282, + "0.8": 256, + "0.9": 212 + }, + "n_dup_removed_abs": { + "0.5": 354, + "0.6": 307, + "0.7": 283, + "0.8": 256, + "0.9": 212 + }, + "largest_dup_abs": { + "0.5": 3, + "0.6": 3, + "0.7": 3, + "0.8": 2, + "0.9": 2 + }, + "babel_mu1_abs": 0.33473145961761475, + "dup_excludes_dead": true + }, + "seed1": { + "d_sae": 6144, + "d_in": 768, + "l0": 128, + "l0_raw": 128.0, + "recon_r2": 0.9484768496878279, + "l1_per_token": 111.2490252923117, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 309, + "dense_rate": 0.05029296875, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 69, + "0.6": 11, + "0.7": 2, + "0.8": 2, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 34, + "0.6": 5, + "0.7": 1, + "0.8": 1, + "0.9": 0 + }, + "n_dup_removed_signed": { + "0.5": 35, + "0.6": 6, + "0.7": 1, + "0.8": 1, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 3, + "0.6": 3, + "0.7": 2, + "0.8": 2, + "0.9": 1 + }, + "babel_mu1_signed": 0.23420310020446777, + "n_dup_abs": { + "0.5": 727, + "0.6": 625, + "0.7": 564, + "0.8": 506, + "0.9": 440 + }, + "n_dup_clusters_abs": { + "0.5": 363, + "0.6": 312, + "0.7": 282, + "0.8": 253, + "0.9": 220 + }, + "n_dup_removed_abs": { + "0.5": 364, + "0.6": 313, + "0.7": 282, + "0.8": 253, + "0.9": 220 + }, + "largest_dup_abs": { + "0.5": 3, + "0.6": 3, + "0.7": 2, + "0.8": 2, + "0.9": 2 + }, + "babel_mu1_abs": 0.3363153040409088, + "dup_excludes_dead": true + }, + "seed2": { + "d_sae": 6144, + "d_in": 768, + "l0": 128, + "l0_raw": 128.0, + "recon_r2": 0.9653791457227991, + "l1_per_token": 111.08873494407304, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 307, + "dense_rate": 0.049967447916666664, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 56, + "0.6": 17, + "0.7": 6, + "0.8": 2, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 27, + "0.6": 8, + "0.7": 3, + "0.8": 1, + "0.9": 0 + }, + "n_dup_removed_signed": { + "0.5": 29, + "0.6": 9, + "0.7": 3, + "0.8": 1, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 3, + "0.6": 3, + "0.7": 2, + "0.8": 2, + "0.9": 1 + }, + "babel_mu1_signed": 0.2326231747865677, + "n_dup_abs": { + "0.5": 713, + "0.6": 616, + "0.7": 556, + "0.8": 498, + "0.9": 424 + }, + "n_dup_clusters_abs": { + "0.5": 355, + "0.6": 307, + "0.7": 278, + "0.8": 249, + "0.9": 212 + }, + "n_dup_removed_abs": { + "0.5": 358, + "0.6": 309, + "0.7": 278, + "0.8": 249, + "0.9": 212 + }, + "largest_dup_abs": { + "0.5": 3, + "0.6": 3, + "0.7": 2, + "0.8": 2, + "0.9": 2 + }, + "babel_mu1_abs": 0.3354438543319702, + "dup_excludes_dead": true + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.6675251126289368, + "dcs_abs": 0.6688010692596436, + "frac_sign_flipped": 0.027180989583333332, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.666540265083313, + "0.6": 0.6674156188964844, + "0.7": 0.6675865054130554, + "0.8": 0.6674880385398865, + "0.9": 0.6675251126289368 + }, + "unique_frac_signed": { + "0.5": 0.99462890625, + "0.6": 0.9991048177083334, + "0.7": 0.999755859375, + "0.8": 0.9998372395833334, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.6629602148314007, + "0.6": 0.6668181602532666, + "0.7": 0.6674235204263823, + "0.8": 0.6673793979086138, + "0.9": 0.6675251126289368 + }, + "matched_frac_signed": { + "0.5": 0.705078125, + "0.6": 0.63818359375, + "0.7": 0.5572916666666666, + "0.8": 0.4513346354166667, + "0.9": 0.2716471354166667 + }, + "matched_cos_abs": { + "0.5": 0.6718766093254089, + "0.6": 0.6714471578598022, + "0.7": 0.6702014803886414, + "0.8": 0.6695786118507385, + "0.9": 0.6692578196525574 + }, + "unique_frac_abs": { + "0.5": 0.9415690104166666, + "0.6": 0.9495442708333334, + "0.7": 0.9540201822916666, + "0.8": 0.9585774739583334, + "0.9": 0.96484375 + }, + "dedup_dcs_abs": { + "0.5": 0.6326181941646306, + "0.6": 0.6375688019131001, + "0.7": 0.6393857384925165, + "0.8": 0.6418429743644083, + "0.9": 0.6457292244303972 + }, + "matched_frac_abs": { + "0.5": 0.6736653645833334, + "0.6": 0.6136067708333334, + "0.7": 0.5367838541666666, + "0.8": 0.4358723958333333, + "0.9": 0.2626953125 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.6676604151725769, + "dcs_abs": 0.6689553260803223, + "frac_sign_flipped": 0.02880859375, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.6664221286773682, + "0.6": 0.6676928997039795, + "0.7": 0.6678035259246826, + "0.8": 0.6676210761070251, + "0.9": 0.6676604151725769 + }, + "unique_frac_signed": { + "0.5": 0.9951171875, + "0.6": 0.9988606770833334, + "0.7": 0.9995930989583334, + "0.8": 0.9998372395833334, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.6631681143771857, + "0.6": 0.6669321818820512, + "0.7": 0.6675317959743552, + "0.8": 0.6675124138225026, + "0.9": 0.6676604151725769 + }, + "matched_frac_signed": { + "0.5": 0.703125, + "0.6": 0.6412760416666666, + "0.7": 0.5558268229166666, + "0.8": 0.4508463541666667, + "0.9": 0.27294921875 + }, + "matched_cos_abs": { + "0.5": 0.67205411195755, + "0.6": 0.6713709235191345, + "0.7": 0.6707805395126343, + "0.8": 0.670266330242157, + "0.9": 0.669474184513092 + }, + "unique_frac_abs": { + "0.5": 0.9420572916666666, + "0.6": 0.9498697916666666, + "0.7": 0.954345703125, + "0.8": 0.9589029947916666, + "0.9": 0.9654947916666666 + }, + "dedup_dcs_abs": { + "0.5": 0.6331134765641764, + "0.6": 0.6377149592541779, + "0.7": 0.6401565256237518, + "0.8": 0.6427203913772246, + "0.9": 0.6463738383026794 + }, + "matched_frac_abs": { + "0.5": 0.671875, + "0.6": 0.6170247395833334, + "0.7": 0.5354817708333334, + "0.8": 0.4348958333333333, + "0.9": 0.26318359375 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.6707451939582825, + "dcs_abs": 0.6720290184020996, + "frac_sign_flipped": 0.02734375, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.6697500348091125, + "0.6": 0.6706778407096863, + "0.7": 0.6708329319953918, + "0.8": 0.6707022786140442, + "0.9": 0.6707451939582825 + }, + "unique_frac_signed": { + "0.5": 0.9947916666666666, + "0.6": 0.998779296875, + "0.7": 0.9996744791666666, + "0.8": 0.9998372395833334, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.666261753377815, + "0.6": 0.6698591421736637, + "0.7": 0.6706145619003412, + "0.8": 0.6705931148317177, + "0.9": 0.6707451939582825 + }, + "matched_frac_signed": { + "0.5": 0.70556640625, + "0.6": 0.64013671875, + "0.7": 0.5556640625, + "0.8": 0.4560546875, + "0.9": 0.2747395833333333 + }, + "matched_cos_abs": { + "0.5": 0.6752935647964478, + "0.6": 0.6748800873756409, + "0.7": 0.6740432381629944, + "0.8": 0.6728955507278442, + "0.9": 0.6722742915153503 + }, + "unique_frac_abs": { + "0.5": 0.9412434895833334, + "0.6": 0.9493815104166666, + "0.7": 0.9544270833333334, + "0.8": 0.9591471354166666, + "0.9": 0.96484375 + }, + "dedup_dcs_abs": { + "0.5": 0.6356156714221773, + "0.6": 0.6407186767028179, + "0.7": 0.6433251218404621, + "0.8": 0.645405839915232, + "0.9": 0.6486396484542638 + }, + "matched_frac_abs": { + "0.5": 0.67529296875, + "0.6": 0.61669921875, + "0.7": 0.5364583333333334, + "0.8": 0.4401041666666667, + "0.9": 0.2644856770833333 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 308, + "n_dense_std": 1.0, + "l0_mean": 128, + "l0_std": 0.0, + "recon_r2_mean": 0.9542054637809102, + "recon_r2_std": 0.009677730918042343, + "l1_per_token_mean": 111.15382592656567, + "l1_per_token_std": 0.08428009842909828, + "babel_mu1_signed_mean": 0.2333275775114695, + "babel_mu1_signed_std": 0.0008037428781287449, + "n_dup_signed_per_threshold": { + "0.5": 62.0, + "0.6": 12.333333333333334, + "0.7": 3.6666666666666665, + "0.8": 2.0, + "0.9": 0.0 + }, + "n_dup_signed_band": 4.5, + "n_dup_signed_band_std": 1.5206906325745548, + "n_dup_clusters_signed_per_threshold": { + "0.5": 30.333333333333332, + "0.6": 5.666666666666667, + "0.7": 1.6666666666666667, + "0.8": 1.0, + "0.9": 0.0 + }, + "n_dup_clusters_signed_band": 2.0833333333333335, + "n_dup_clusters_signed_band_std": 0.8036375634160795, + "largest_dup_signed_per_threshold": { + "0.5": 3.0, + "0.6": 3.0, + "0.7": 2.3333333333333335, + "0.8": 2.0, + "0.9": 1.0 + }, + "largest_dup_signed_band": 2.0833333333333335, + "largest_dup_signed_band_std": 0.14433756729740643, + "dcs_signed_mean": 0.668643573919932, + "dcs_signed_std": 0.001821313203097122, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.6641300275288005, + "0.6": 0.6678698281029938, + "0.7": 0.6685232927670262, + "0.8": 0.6684949755209447, + "0.9": 0.668643573919932 + }, + "dedup_dcs_signed_band": 0.6683829175777242, + "dedup_dcs_signed_band_std": 0.0017937953918082802, + "matched_cos_signed_per_threshold": { + "0.5": 0.6675708095232645, + "0.6": 0.6685954531033834, + "0.7": 0.66874098777771, + "0.8": 0.668603797753652, + "0.9": 0.668643573919932 + }, + "matched_cos_signed_band": 0.6686459531386694, + "matched_cos_signed_band_std": 0.0018156222772486765, + "unique_frac_signed_per_threshold": { + "0.5": 0.9948459201388888, + "0.6": 0.9989149305555557, + "0.7": 0.9996744791666666, + "0.8": 0.9998372395833334, + "0.9": 1.0 + }, + "unique_frac_signed_band": 0.999606662326389, + "unique_frac_signed_band_std": 5.873110648499001e-05, + "matched_frac_signed_per_threshold": { + "0.5": 0.70458984375, + "0.6": 0.6398654513888888, + "0.7": 0.5562608506944444, + "0.8": 0.4527452256944445, + "0.9": 0.2731119791666667 + }, + "matched_frac_signed_band": 0.4804958767361111, + "matched_frac_signed_band_std": 0.0010440269855645414, + "babel_mu1_abs_mean": 0.3354968726634979, + "babel_mu1_abs_std": 0.0007932521647675573, + "n_dup_abs_per_threshold": { + "0.5": 715.3333333333334, + "0.6": 618.0, + "0.7": 561.6666666666666, + "0.8": 505.3333333333333, + "0.9": 429.3333333333333 + }, + "n_dup_abs_band": 528.5833333333333, + "n_dup_abs_band_std": 5.125508104893926, + "n_dup_clusters_abs_per_threshold": { + "0.5": 356.6666666666667, + "0.6": 308.3333333333333, + "0.7": 280.6666666666667, + "0.8": 252.66666666666666, + "0.9": 214.66666666666666 + }, + "n_dup_clusters_abs_band": 264.0833333333333, + "n_dup_clusters_abs_band_std": 2.6259918760981216, + "largest_dup_abs_per_threshold": { + "0.5": 3.0, + "0.6": 3.0, + "0.7": 2.3333333333333335, + "0.8": 2.0, + "0.9": 2.0 + }, + "largest_dup_abs_band": 2.3333333333333335, + "largest_dup_abs_band_std": 0.14433756729740643, + "dcs_abs_mean": 0.6699284712473551, + "dcs_abs_std": 0.0018207615312446272, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.6337824473836614, + "0.6": 0.638667479290032, + "0.7": 0.6409557953189101, + "0.8": 0.6433230685522883, + "0.9": 0.6469142370624468 + }, + "dedup_dcs_abs_band": 0.6424651450559193, + "dedup_dcs_abs_band_std": 0.0018074647391084575, + "matched_cos_abs_per_threshold": { + "0.5": 0.6730747620264689, + "0.6": 0.6725660562515259, + "0.7": 0.6716750860214233, + "0.8": 0.6709134976069132, + "0.9": 0.6703354318936666 + }, + "matched_cos_abs_band": 0.6713725179433823, + "matched_cos_abs_band_std": 0.0018709087573115922, + "unique_frac_abs_per_threshold": { + "0.5": 0.9416232638888888, + "0.6": 0.9495985243055555, + "0.7": 0.9542643229166666, + "0.8": 0.9588758680555555, + "0.9": 0.9650607638888888 + }, + "unique_frac_abs_band": 0.9569498697916666, + "unique_frac_abs_band_std": 0.00020345052083325932, + "matched_frac_abs_per_threshold": { + "0.5": 0.6736111111111112, + "0.6": 0.6157769097222222, + "0.7": 0.5362413194444445, + "0.8": 0.43695746527777773, + "0.9": 0.2634548611111111 + }, + "matched_frac_abs_band": 0.4631076388888889, + "matched_frac_abs_band_std": 0.00116897034648523, + "alive_source": "stats", + "frac_denominator": "d_sae", + "d_sae": 6144 + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/seed0/config.json b/lm/pythia-160m_layer8/topk/x8/l0_128/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..20071d712b97551e507642401ad3a55036610413 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/seed0/config.json @@ -0,0 +1,12 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "topk", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 128, + "top_k_inference": 0, + "class_type": "TopKSAE", + "module_name": "consdl.models.topk" +} diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..ef129f02849ffb99d41c516ccb8e2fd264af6c9e --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 128.0, + "l0_per_token_0.05": 128.0, + "l0_per_token_0.1": 127.9973541015625, + "l0_per_token_0.5": 76.05146728515625, + "l0_per_token_1.0": 25.252570654296875, + "l1_per_token": 111.12371754331225, + "r2_score": 0.9487603959321035, + "l2_loss": 0.03411384182129974 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/seed1/config.json b/lm/pythia-160m_layer8/topk/x8/l0_128/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..556ec95876818cad5126cb5a7fb3e52b5c9b02f6 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/seed1/config.json @@ -0,0 +1,12 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "topk", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 128, + "top_k_inference": 0, + "class_type": "TopKSAE", + "module_name": "consdl.models.topk" +} diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..610620b75253e35f57b81e42098c342f2c2893ac --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 128.0, + "l0_per_token_0.05": 128.0, + "l0_per_token_0.1": 127.9993041015625, + "l0_per_token_0.5": 76.3709333984375, + "l0_per_token_1.0": 25.206501416015627, + "l1_per_token": 111.2490252923117, + "r2_score": 0.9484768496878279, + "l2_loss": 0.034302618684474255 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/seed2/config.json b/lm/pythia-160m_layer8/topk/x8/l0_128/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c633eec338943c277944540dbafcbdd6750fb4af --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/seed2/config.json @@ -0,0 +1,12 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "topk", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 128, + "top_k_inference": 0, + "class_type": "TopKSAE", + "module_name": "consdl.models.topk" +} diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..d0ca07ef7cebf028002d0e70a87f4cd56216b6c3 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 128.0, + "l0_per_token_0.001": 128.0, + "l0_per_token_0.01": 128.0, + "l0_per_token_0.05": 127.99999995117187, + "l0_per_token_0.1": 127.99753334960937, + "l0_per_token_0.5": 75.98884814453125, + "l0_per_token_1.0": 25.1584048828125, + "l1_per_token": 111.08873494407304, + "r2_score": 0.9653791457227991, + "l2_loss": 0.03408511153066325 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_128/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/code_metrics_step_final.json b/lm/pythia-160m_layer8/topk/x8/l0_32/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..48c5c091b76f6754ee42c51c4a16dd5351e14693 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 32.0, + "recon_r2": 0.9052632670944003, + "l1_per_token": 57.8659340379438 + }, + "seed1": { + "l0": 32.0, + "recon_r2": 0.9050926654595287, + "l1_per_token": 57.89805520942179 + }, + "seed2": { + "l0": 32.0, + "recon_r2": 0.9051852319138651, + "l1_per_token": 57.8361864238528 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.7432986563426907, + "mcc_abs": 0.743311620686602, + "act_iou": 0.4454948553695755, + "act_iou_n_pairs": 6144, + "input_r2": 0.9670196169961698, + "matched_mcc_signed": { + "0.6": 0.8428682011266775, + "0.7": 0.8691962149948246, + "0.8": 0.8965139618946378, + "0.9": 0.9259628334761729 + }, + "matched_mcc_abs": { + "0.6": 0.8431333416562056, + "0.7": 0.8690441738539502, + "0.8": 0.8965318097267021, + "0.9": 0.9258875045033148 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7365748028971023, + "mcc_abs": 0.7365864533058785, + "act_iou": 0.43851382930958754, + "act_iou_n_pairs": 6144, + "input_r2": 0.9662396954037573, + "matched_mcc_signed": { + "0.6": 0.840615531057578, + "0.7": 0.8671233170295205, + "0.8": 0.8966889842965253, + "0.9": 0.9263378962805466 + }, + "matched_mcc_abs": { + "0.6": 0.83986003762432, + "0.7": 0.8668526880501591, + "0.8": 0.8963804614627346, + "0.9": 0.9262039911364368 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.7365251543910428, + "mcc_abs": 0.7365402100607006, + "act_iou": 0.4382382058124839, + "act_iou_n_pairs": 6144, + "input_r2": 0.966263709824976, + "matched_mcc_signed": { + "0.6": 0.8393754632367143, + "0.7": 0.8675013899659789, + "0.8": 0.8945507035287771, + "0.9": 0.9264502798172064 + }, + "matched_mcc_abs": { + "0.6": 0.8390117984718679, + "0.7": 0.8676742153607105, + "0.8": 0.8944931522487517, + "0.9": 0.9263308416417707 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 32.0, + "l0_std": 0.0, + "recon_r2_mean": 0.9051803881559314, + "recon_r2_std": 8.540389890634721e-05, + "l1_per_token_mean": 57.86672522373947, + "l1_per_token_std": 0.03094198019144633, + "input_r2_mean": 0.9665076740749677, + "input_r2_std": 0.0004435181383363134, + "mcc_signed_mean": 0.7387995378769453, + "mcc_signed_std": 0.003896429964742965, + "mcc_abs_mean": 0.7388127613510603, + "mcc_abs_std": 0.0038961950798405373, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8409530651403232, + "0.7": 0.867940307330108, + "0.8": 0.8959178832399801, + "0.9": 0.926250336524642 + }, + "matched_mcc_signed_band": 0.8827653980587633, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8406683925841311, + "0.7": 0.86785702575494, + "0.8": 0.8958018078127296, + "0.9": 0.9261407790938407 + }, + "matched_mcc_abs_band": 0.8826170013114103, + "act_iou_mean": 0.44074896349721565, + "act_iou_std": 0.004112372712454137, + "act_iou_n_pairs": 6144.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/dict_metrics_step_final.json b/lm/pythia-160m_layer8/topk/x8/l0_32/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..931de05aa55f8292ac7867fa307d8db189633924 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/dict_metrics_step_final.json @@ -0,0 +1,545 @@ +{ + "schema": 2, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 6144, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.9057426249388189, + "l1_per_token": 57.7649642546186, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 18, + "dense_rate": 0.0029296875, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 554, + "0.6": 158, + "0.7": 39, + "0.8": 12, + "0.9": 2 + }, + "n_dup_clusters_signed": { + "0.5": 263, + "0.6": 72, + "0.7": 16, + "0.8": 4, + "0.9": 1 + }, + "largest_dup_signed": { + "0.5": 10, + "0.6": 9, + "0.7": 8, + "0.8": 6, + "0.9": 2 + }, + "babel_mu1_signed": 0.36380115151405334, + "n_dup_abs": { + "0.5": 677, + "0.6": 244, + "0.7": 111, + "0.8": 62, + "0.9": 22 + }, + "n_dup_clusters_abs": { + "0.5": 323, + "0.6": 115, + "0.7": 52, + "0.8": 29, + "0.9": 11 + }, + "largest_dup_abs": { + "0.5": 10, + "0.6": 9, + "0.7": 8, + "0.8": 6, + "0.9": 2 + }, + "babel_mu1_abs": 0.37708500027656555 + }, + "seed1": { + "d_sae": 6144, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.9056473311257675, + "l1_per_token": 57.79262024250857, + "stats_source": "stats", + "n_dead": 1, + "dead_rate": 0.00016276041666666666, + "alive_source": "stats", + "n_dense": 18, + "dense_rate": 0.0029296875, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 543, + "0.6": 138, + "0.7": 38, + "0.8": 10, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 259, + "0.6": 63, + "0.7": 15, + "0.8": 2, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 12, + "0.6": 9, + "0.7": 9, + "0.8": 8, + "0.9": 1 + }, + "babel_mu1_signed": 0.365743488073349, + "n_dup_abs": { + "0.5": 662, + "0.6": 234, + "0.7": 110, + "0.8": 62, + "0.9": 26 + }, + "n_dup_clusters_abs": { + "0.5": 318, + "0.6": 111, + "0.7": 51, + "0.8": 28, + "0.9": 13 + }, + "largest_dup_abs": { + "0.5": 12, + "0.6": 9, + "0.7": 9, + "0.8": 8, + "0.9": 2 + }, + "babel_mu1_abs": 0.3791674077510834 + }, + "seed2": { + "d_sae": 6144, + "d_in": 768, + "l0": 32, + "l0_raw": 32.0, + "recon_r2": 0.9057064882874792, + "l1_per_token": 57.72785167534503, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 20, + "dense_rate": 0.0032552083333333335, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 571, + "0.6": 163, + "0.7": 40, + "0.8": 8, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 269, + "0.6": 74, + "0.7": 16, + "0.8": 2, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 14, + "0.6": 12, + "0.7": 9, + "0.8": 6, + "0.9": 1 + }, + "babel_mu1_signed": 0.3644007742404938, + "n_dup_abs": { + "0.5": 684, + "0.6": 253, + "0.7": 108, + "0.8": 58, + "0.9": 24 + }, + "n_dup_clusters_abs": { + "0.5": 325, + "0.6": 119, + "0.7": 50, + "0.8": 27, + "0.9": 12 + }, + "largest_dup_abs": { + "0.5": 14, + "0.6": 12, + "0.7": 9, + "0.8": 6, + "0.9": 2 + }, + "babel_mu1_abs": 0.37805747985839844 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.8138559460639954, + "dcs_abs": 0.8145877718925476, + "frac_sign_flipped": 0.009440104166666666, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6143, + "matched_cos_signed": { + "0.5": 0.8070195913314819, + "0.6": 0.8122165203094482, + "0.7": 0.8135913014411926, + "0.8": 0.8137865662574768, + "0.9": 0.8139371275901794 + }, + "unique_frac_signed": { + "0.5": 0.9532025718238789, + "0.6": 0.9868967201106861, + "0.7": 0.9962562057459103, + "0.8": 0.9986978106942297, + "0.9": 0.9999186131683894 + }, + "dedup_dcs_signed": { + "0.5": 0.7692531499694243, + "0.6": 0.8015738199131089, + "0.7": 0.8105453830016798, + "0.8": 0.8127268620937168, + "0.9": 0.8138708838262346 + }, + "matched_frac_signed": { + "0.5": 0.8277040774802636, + "0.6": 0.8202164889720843, + "0.7": 0.7748026369333442, + "0.8": 0.6939041263123626, + "0.9": 0.5487100187189713 + }, + "matched_cos_abs": { + "0.5": 0.8090957999229431, + "0.6": 0.8131924271583557, + "0.7": 0.814351499080658, + "0.8": 0.8144892454147339, + "0.9": 0.8145724534988403 + }, + "unique_frac_abs": { + "0.5": 0.9431919915357695, + "0.6": 0.9794905184341174, + "0.7": 0.9903963538699438, + "0.8": 0.9945470822820868, + "0.9": 0.9980467160413445 + }, + "dedup_dcs_abs": { + "0.5": 0.7631326788725472, + "0.6": 0.796514272064036, + "0.7": 0.8065307554580066, + "0.8": 0.8100479025773621, + "0.9": 0.8129813621922585 + }, + "matched_frac_abs": { + "0.5": 0.8192398469927565, + "0.6": 0.8159843737283308, + "0.7": 0.7715471636689184, + "0.8": 0.6919508423537072, + "0.9": 0.547407829413201 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.8071193099021912, + "dcs_abs": 0.8078550696372986, + "frac_sign_flipped": 0.009602864583333334, + "alive_source": "stats", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.8007305264472961, + "0.6": 0.804703414440155, + "0.7": 0.8067680597305298, + "0.8": 0.8072454929351807, + "0.9": 0.8072338104248047 + }, + "unique_frac_signed": { + "0.5": 0.9517415364583334, + "0.6": 0.9857584635416666, + "0.7": 0.9961751302083334, + "0.8": 0.9988606770833334, + "0.9": 0.9999186197916666 + }, + "dedup_dcs_signed": { + "0.5": 0.7620885015300398, + "0.6": 0.7932432014252602, + "0.7": 0.803682276949985, + "0.8": 0.8063257796457037, + "0.9": 0.8071681175691385 + }, + "matched_frac_signed": { + "0.5": 0.8204752604166666, + "0.6": 0.8094075520833334, + "0.7": 0.7648111979166666, + "0.8": 0.6748046875, + "0.9": 0.52734375 + }, + "matched_cos_abs": { + "0.5": 0.8020846247673035, + "0.6": 0.8061532974243164, + "0.7": 0.8076242208480835, + "0.8": 0.8080651164054871, + "0.9": 0.8079200387001038 + }, + "unique_frac_abs": { + "0.5": 0.9419759114583334, + "0.6": 0.9785970052083334, + "0.7": 0.990478515625, + "0.8": 0.9947916666666666, + "0.9": 0.9981282552083334 + }, + "dedup_dcs_abs": { + "0.5": 0.755544395481896, + "0.6": 0.7888992025982589, + "0.7": 0.7999344394484069, + "0.8": 0.8038564439242084, + "0.9": 0.8064078185755837 + }, + "matched_frac_abs": { + "0.5": 0.8141276041666666, + "0.6": 0.8059895833333334, + "0.7": 0.76220703125, + "0.8": 0.6741536458333334, + "0.9": 0.5262044270833334 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.806806743144989, + "dcs_abs": 0.8078908324241638, + "frac_sign_flipped": 0.012044270833333334, + "alive_source": "stats", + "n_alive_a": 6143, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.8005146384239197, + "0.6": 0.8049227595329285, + "0.7": 0.8062835335731506, + "0.8": 0.806787371635437, + "0.9": 0.806902289390564 + }, + "unique_frac_signed": { + "0.5": 0.9523073166761618, + "0.6": 0.9866525596158542, + "0.7": 0.9961748189142997, + "0.8": 0.998860584357451, + "0.9": 1.0 + }, + "dedup_dcs_signed": { + "0.5": 0.7623359472774708, + "0.6": 0.7941791009862206, + "0.7": 0.803199353050815, + "0.8": 0.8058681054839846, + "0.9": 0.806902289390564 + }, + "matched_frac_signed": { + "0.5": 0.8207048099617482, + "0.6": 0.8094734271994791, + "0.7": 0.7585252706112151, + "0.8": 0.6769756653373484, + "0.9": 0.5303165947749654 + }, + "matched_cos_abs": { + "0.5": 0.8021262288093567, + "0.6": 0.8062965273857117, + "0.7": 0.807614803314209, + "0.8": 0.8079354763031006, + "0.9": 0.8079301714897156 + }, + "unique_frac_abs": { + "0.5": 0.9427850573777163, + "0.6": 0.9790835842760641, + "0.7": 0.9904777407015545, + "0.8": 0.994709855945308, + "0.9": 0.9979653292097339 + }, + "dedup_dcs_abs": { + "0.5": 0.7562326226522006, + "0.6": 0.7894316940221463, + "0.7": 0.799924485743788, + "0.8": 0.8036613812465611, + "0.9": 0.8062862995692108 + }, + "matched_frac_abs": { + "0.5": 0.8124033531374624, + "0.6": 0.8054040856189468, + "0.7": 0.7555953446732319, + "0.8": 0.6751851550419142, + "0.9": 0.5293399527956376 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.5773502691896258, + "n_dense_mean": 19, + "n_dense_std": 1.1547005383792515, + "l0_mean": 32, + "l0_std": 0.0, + "recon_r2_mean": 0.9056988147840218, + "recon_r2_std": null, + "l1_per_token_mean": 57.76181205749074, + "l1_per_token_std": 0.03249915529355073, + "babel_mu1_signed_mean": 0.3646484712759654, + "babel_mu1_signed_std": 0.000994576891663075, + "n_dup_signed_per_threshold": { + "0.5": 556.0, + "0.6": 153.0, + "0.7": 39.0, + "0.8": 10.0, + "0.9": 0.6666666666666666 + }, + "n_dup_signed_band": 50.666666666666664, + "n_dup_signed_band_std": 3.608439182435161, + "n_dup_clusters_signed_per_threshold": { + "0.5": 263.6666666666667, + "0.6": 69.66666666666667, + "0.7": 15.666666666666666, + "0.8": 2.6666666666666665, + "0.9": 0.3333333333333333 + }, + "n_dup_clusters_signed_band": 22.083333333333336, + "n_dup_clusters_signed_band_std": 1.808544534517559, + "largest_dup_signed_per_threshold": { + "0.5": 12.0, + "0.6": 10.0, + "0.7": 8.666666666666666, + "0.8": 6.666666666666667, + "0.9": 1.3333333333333333 + }, + "largest_dup_signed_band": 6.666666666666666, + "largest_dup_signed_band_std": 0.38188130791298663, + "dcs_signed_mean": 0.8092606663703918, + "dcs_signed_std": 0.003982696459844835, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.7645591995923117, + "0.6": 0.7963320407748632, + "0.7": 0.80580900433416, + "0.8": 0.808306915741135, + "0.9": 0.8093137635953124 + }, + "dedup_dcs_signed_band": 0.8049404311113676, + "dedup_dcs_signed_band_std": 0.004104065780620972, + "matched_cos_signed_per_threshold": { + "0.5": 0.8027549187342325, + "0.6": 0.8072808980941772, + "0.7": 0.8088809649149576, + "0.8": 0.8092731436093649, + "0.9": 0.809357742468516 + }, + "matched_cos_signed_band": 0.8086981872717539, + "matched_cos_signed_band_std": 0.004059203977334967, + "unique_frac_signed_per_threshold": { + "0.5": 0.9524171416527913, + "0.6": 0.9864359144227356, + "0.7": 0.9962020516228479, + "0.8": 0.998806357378338, + "0.9": 0.9999457443200187 + }, + "unique_frac_signed_band": 0.9953475169359851, + "unique_frac_signed_band_std": 0.0001469656830483683, + "matched_frac_signed_per_threshold": { + "0.5": 0.8229613826195594, + "0.6": 0.813032489418299, + "0.7": 0.7660463684870753, + "0.8": 0.6818948263832371, + "0.9": 0.5354567878313122 + }, + "matched_frac_signed_band": 0.6991076180299809, + "matched_frac_signed_band_std": 0.008921681947232774, + "babel_mu1_abs_mean": 0.3781032959620158, + "babel_mu1_abs_std": 0.0010419594804957294, + "n_dup_abs_per_threshold": { + "0.5": 674.3333333333334, + "0.6": 243.66666666666666, + "0.7": 109.66666666666667, + "0.8": 60.666666666666664, + "0.9": 24.0 + }, + "n_dup_abs_band": 109.5, + "n_dup_abs_band_std": 1.3919410907075054, + "n_dup_clusters_abs_per_threshold": { + "0.5": 322.0, + "0.6": 115.0, + "0.7": 51.0, + "0.8": 28.0, + "0.9": 12.0 + }, + "n_dup_clusters_abs_band": 51.5, + "n_dup_clusters_abs_band_std": 0.6614378277661477, + "largest_dup_abs_per_threshold": { + "0.5": 12.0, + "0.6": 10.0, + "0.7": 8.666666666666666, + "0.8": 6.666666666666667, + "0.9": 2.0 + }, + "largest_dup_abs_band": 6.833333333333333, + "largest_dup_abs_band_std": 0.5204164998665333, + "dcs_abs_mean": 0.8101112246513367, + "dcs_abs_std": 0.0038768448700362045, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.758303232335548, + "0.6": 0.7916150562281471, + "0.7": 0.8021298935500671, + "0.8": 0.8058552425827106, + "0.9": 0.8085584934456843 + }, + "dedup_dcs_abs_band": 0.8020396714516523, + "dedup_dcs_abs_band_std": 0.0038789280193405372, + "matched_cos_abs_per_threshold": { + "0.5": 0.8044355511665344, + "0.6": 0.8085474173227946, + "0.7": 0.8098635077476501, + "0.8": 0.8101632793744405, + "0.9": 0.8101408878962199 + }, + "matched_cos_abs_band": 0.8096787730852762, + "matched_cos_abs_band_std": 0.003873414388237887, + "unique_frac_abs_per_threshold": { + "0.5": 0.9426509867906064, + "0.6": 0.9790570359728382, + "0.7": 0.9904508700654994, + "0.8": 0.9946828682980206, + "0.9": 0.998046766819804 + }, + "unique_frac_abs_band": 0.9905593852890405, + "unique_frac_abs_band_std": 6.065390065771041e-05, + "matched_frac_abs_per_threshold": { + "0.5": 0.8152569347656285, + "0.6": 0.8091260142268704, + "0.7": 0.7631165131973834, + "0.8": 0.6804298810763183, + "0.9": 0.5343174030973907 + }, + "matched_frac_abs_band": 0.6967474528994907, + "matched_frac_abs_band_std": 0.008646989164431814, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/seed0/config.json b/lm/pythia-160m_layer8/topk/x8/l0_32/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3ce12067a8863b91a9d79c7c4c7ea13fbc128cee --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/seed0/config.json @@ -0,0 +1,12 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "topk", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 32, + "top_k_inference": 0, + "class_type": "TopKSAE", + "module_name": "consdl.models.topk" +} diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..fe7568bcda2c4643bcba91de24155ab2b3231774 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 32.0, + "l0_per_token_0.05": 32.0, + "l0_per_token_0.1": 31.999999853515625, + "l0_per_token_0.5": 31.550620947265624, + "l0_per_token_1.0": 18.63238134765625, + "l1_per_token": 57.7649642546186, + "r2_score": 0.9057426249388189, + "l2_loss": 0.06275382571393973 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/seed1/config.json b/lm/pythia-160m_layer8/topk/x8/l0_32/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..7964558bf0dbf0fe06b1ae7d224de8b79e6126db --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/seed1/config.json @@ -0,0 +1,12 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "topk", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 32, + "top_k_inference": 0, + "class_type": "TopKSAE", + "module_name": "consdl.models.topk" +} diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..ce8ceb235fc0218cb4a893cee9c5d31a78400ab6 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 32.0, + "l0_per_token_0.05": 32.0, + "l0_per_token_0.1": 31.99999833984375, + "l0_per_token_0.5": 31.566606298828123, + "l0_per_token_1.0": 18.645304931640624, + "l1_per_token": 57.79262024250857, + "r2_score": 0.9056473311257675, + "l2_loss": 0.06281726956999835 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/seed2/config.json b/lm/pythia-160m_layer8/topk/x8/l0_32/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..09e46fd324567f67c3c939dc55a6795e157259cf --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/seed2/config.json @@ -0,0 +1,12 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "topk", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 32, + "top_k_inference": 0, + "class_type": "TopKSAE", + "module_name": "consdl.models.topk" +} diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..c5237f988f1d532590b05e8b35785b814f3a6116 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 32.0, + "l0_per_token_0.001": 32.0, + "l0_per_token_0.01": 32.0, + "l0_per_token_0.05": 32.0, + "l0_per_token_0.1": 31.9999955078125, + "l0_per_token_0.5": 31.52871025390625, + "l0_per_token_1.0": 18.545965869140623, + "l1_per_token": 57.72785167534503, + "r2_score": 0.9057064882874792, + "l2_loss": 0.06277788444800253 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_32/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/code_metrics_step_final.json b/lm/pythia-160m_layer8/topk/x8/l0_64/code_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..a1aba4c8a95834484d5408927871e9743e4368ab --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/code_metrics_step_final.json @@ -0,0 +1,136 @@ +{ + "schema": 1, + "step": "final", + "n_tokens": 600000, + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "l0": 64.0, + "recon_r2": 0.9267746828725079, + "l1_per_token": 77.27601765746131 + }, + "seed1": { + "l0": 64.0, + "recon_r2": 0.9264271047492493, + "l1_per_token": 77.27067061329068 + }, + "seed2": { + "l0": 64.0, + "recon_r2": 0.9268145398989455, + "l1_per_token": 76.94912996633016 + } + }, + "per_pair": { + "seed0\u2194seed1": { + "alive_source": "stats", + "mcc_signed": 0.6960555292807459, + "mcc_abs": 0.6960705346884507, + "act_iou": 0.3839193305338265, + "act_iou_n_pairs": 6144, + "input_r2": 0.9661040482585324, + "matched_mcc_signed": { + "0.6": 0.8404351606909669, + "0.7": 0.8705856417157846, + "0.8": 0.897436628021344, + "0.9": 0.9297044912547116 + }, + "matched_mcc_abs": { + "0.6": 0.8409195427682767, + "0.7": 0.8712317393089912, + "0.8": 0.8974447151919426, + "0.9": 0.9298313251108274 + } + }, + "seed0\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.6971133519331362, + "mcc_abs": 0.697127166120462, + "act_iou": 0.3853859205304966, + "act_iou_n_pairs": 6144, + "input_r2": 0.966501947079531, + "matched_mcc_signed": { + "0.6": 0.8406502515780893, + "0.7": 0.8699897797853462, + "0.8": 0.8982586740470886, + "0.9": 0.9323025193781681 + }, + "matched_mcc_abs": { + "0.6": 0.841277672807065, + "0.7": 0.8702129045717008, + "0.8": 0.8985986748380633, + "0.9": 0.9322789272242995 + } + }, + "seed1\u2194seed2": { + "alive_source": "stats", + "mcc_signed": 0.6997779915406118, + "mcc_abs": 0.6997894510159854, + "act_iou": 0.38876197687313113, + "act_iou_n_pairs": 6144, + "input_r2": 0.9668359607542617, + "matched_mcc_signed": { + "0.6": 0.8432383271035775, + "0.7": 0.8712309413511657, + "0.8": 0.8990130105547203, + "0.9": 0.9281427495224437 + }, + "matched_mcc_abs": { + "0.6": 0.8437787818518601, + "0.7": 0.8718386757988741, + "0.8": 0.8996147035671682, + "0.9": 0.9285388475238925 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "l0_mean": 64.0, + "l0_std": 0.0, + "recon_r2_mean": 0.9266721091735675, + "recon_r2_std": 0.00021311387001026722, + "l1_per_token_mean": 77.16527274569405, + "l1_per_token_std": 0.18720422944928422, + "input_r2_mean": 0.9664806520307749, + "input_r2_std": 0.0003664206389893036, + "mcc_signed_mean": 0.6976489575848314, + "mcc_signed_std": 0.001918159633510043, + "mcc_abs_mean": 0.6976623839416328, + "mcc_abs_std": 0.0019163580693271195, + "matched_mcc_signed_per_threshold": { + "0.6": 0.8414412464575446, + "0.7": 0.8706021209507654, + "0.8": 0.8982361042077177, + "0.9": 0.9300499200517743 + }, + "matched_mcc_signed_band": 0.8850823479169505, + "matched_mcc_abs_per_threshold": { + "0.6": 0.8419919991424006, + "0.7": 0.8710944398931887, + "0.8": 0.8985526978657248, + "0.9": 0.9302163666196731 + }, + "matched_mcc_abs_band": 0.8854638758802468, + "act_iou_mean": 0.3860224093124847, + "act_iou_std": 0.002483272914742305, + "act_iou_n_pairs": 6144.0, + "alive_source": "stats" + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/dict_metrics_step_final.json b/lm/pythia-160m_layer8/topk/x8/l0_64/dict_metrics_step_final.json new file mode 100644 index 0000000000000000000000000000000000000000..fdb3df697e689b34a53c74bc5fe4618a62d1eee5 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/dict_metrics_step_final.json @@ -0,0 +1,595 @@ +{ + "schema": 3, + "step": "final", + "linkage": "complete", + "thresholds": [ + 0.5, + 0.6, + 0.7, + 0.8, + 0.9 + ], + "seeds": [ + "seed0", + "seed1", + "seed2" + ], + "per_seed": { + "seed0": { + "d_sae": 6144, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.9271624524349621, + "l1_per_token": 77.13320429374109, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 74, + "dense_rate": 0.012044270833333334, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 168, + "0.6": 40, + "0.7": 10, + "0.8": 8, + "0.9": 3 + }, + "n_dup_clusters_signed": { + "0.5": 80, + "0.6": 17, + "0.7": 2, + "0.8": 2, + "0.9": 1 + }, + "n_dup_removed_signed": { + "0.5": 88, + "0.6": 23, + "0.7": 8, + "0.8": 6, + "0.9": 2 + }, + "largest_dup_signed": { + "0.5": 9, + "0.6": 8, + "0.7": 8, + "0.8": 6, + "0.9": 3 + }, + "babel_mu1_signed": 0.287741482257843, + "n_dup_abs": { + "0.5": 461, + "0.6": 301, + "0.7": 236, + "0.8": 210, + "0.9": 149 + }, + "n_dup_clusters_abs": { + "0.5": 225, + "0.6": 147, + "0.7": 115, + "0.8": 103, + "0.9": 74 + }, + "n_dup_removed_abs": { + "0.5": 236, + "0.6": 154, + "0.7": 121, + "0.8": 107, + "0.9": 75 + }, + "largest_dup_abs": { + "0.5": 9, + "0.6": 8, + "0.7": 8, + "0.8": 6, + "0.9": 3 + }, + "babel_mu1_abs": 0.32980963587760925, + "dup_excludes_dead": true + }, + "seed1": { + "d_sae": 6144, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.9268578212374675, + "l1_per_token": 77.10306360961033, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 71, + "dense_rate": 0.011555989583333334, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 171, + "0.6": 46, + "0.7": 18, + "0.8": 5, + "0.9": 2 + }, + "n_dup_clusters_signed": { + "0.5": 81, + "0.6": 21, + "0.7": 7, + "0.8": 1, + "0.9": 1 + }, + "n_dup_removed_signed": { + "0.5": 90, + "0.6": 25, + "0.7": 11, + "0.8": 4, + "0.9": 1 + }, + "largest_dup_signed": { + "0.5": 6, + "0.6": 6, + "0.7": 6, + "0.8": 5, + "0.9": 2 + }, + "babel_mu1_signed": 0.289554625749588, + "n_dup_abs": { + "0.5": 483, + "0.6": 301, + "0.7": 239, + "0.8": 193, + "0.9": 150 + }, + "n_dup_clusters_abs": { + "0.5": 236, + "0.6": 148, + "0.7": 117, + "0.8": 95, + "0.9": 75 + }, + "n_dup_removed_abs": { + "0.5": 247, + "0.6": 153, + "0.7": 122, + "0.8": 98, + "0.9": 75 + }, + "largest_dup_abs": { + "0.5": 7, + "0.6": 7, + "0.7": 7, + "0.8": 5, + "0.9": 2 + }, + "babel_mu1_abs": 0.33156490325927734, + "dup_excludes_dead": true + }, + "seed2": { + "d_sae": 6144, + "d_in": 768, + "l0": 64, + "l0_raw": 64.0, + "recon_r2": 0.9507906465203091, + "l1_per_token": 76.79829959712016, + "stats_source": "stats", + "n_dead": 0, + "dead_rate": 0.0, + "alive_source": "stats", + "n_dense": 74, + "dense_rate": 0.012044270833333334, + "dense_threshold": 0.1, + "n_dup_signed": { + "0.5": 167, + "0.6": 38, + "0.7": 13, + "0.8": 7, + "0.9": 0 + }, + "n_dup_clusters_signed": { + "0.5": 78, + "0.6": 15, + "0.7": 4, + "0.8": 2, + "0.9": 0 + }, + "n_dup_removed_signed": { + "0.5": 89, + "0.6": 23, + "0.7": 9, + "0.8": 5, + "0.9": 0 + }, + "largest_dup_signed": { + "0.5": 9, + "0.6": 9, + "0.7": 5, + "0.8": 4, + "0.9": 1 + }, + "babel_mu1_signed": 0.2887057363986969, + "n_dup_abs": { + "0.5": 475, + "0.6": 290, + "0.7": 233, + "0.8": 187, + "0.9": 146 + }, + "n_dup_clusters_abs": { + "0.5": 232, + "0.6": 141, + "0.7": 114, + "0.8": 92, + "0.9": 73 + }, + "n_dup_removed_abs": { + "0.5": 243, + "0.6": 149, + "0.7": 119, + "0.8": 95, + "0.9": 73 + }, + "largest_dup_abs": { + "0.5": 9, + "0.6": 9, + "0.7": 5, + "0.8": 4, + "0.9": 2 + }, + "babel_mu1_abs": 0.3307194411754608, + "dup_excludes_dead": true + } + }, + "per_pair": { + "seed0\u2194seed1": { + "dcs_signed": 0.7521495819091797, + "dcs_abs": 0.7530983090400696, + "frac_sign_flipped": 0.016927083333333332, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.7492056488990784, + "0.6": 0.7515338063240051, + "0.7": 0.7524080276489258, + "0.8": 0.7522502541542053, + "0.9": 0.7522275447845459 + }, + "unique_frac_signed": { + "0.5": 0.9855143229166666, + "0.6": 0.99609375, + "0.7": 0.9984537760416666, + "0.8": 0.9991861979166666, + "0.9": 0.999755859375 + }, + "dedup_dcs_signed": { + "0.5": 0.738352897800117, + "0.6": 0.748598127393052, + "0.7": 0.7512446363301326, + "0.8": 0.7516380713301866, + "0.9": 0.75204389548162 + }, + "matched_frac_signed": { + "0.5": 0.7955729166666666, + "0.6": 0.7459309895833334, + "0.7": 0.6746419270833334, + "0.8": 0.5821940104166666, + "0.9": 0.4191080729166667 + }, + "matched_cos_abs": { + "0.5": 0.7530577182769775, + "0.6": 0.7540024518966675, + "0.7": 0.7542683482170105, + "0.8": 0.7541047930717468, + "0.9": 0.7540658712387085 + }, + "unique_frac_abs": { + "0.5": 0.960693359375, + "0.6": 0.9750162760416666, + "0.7": 0.980224609375, + "0.8": 0.9833170572916666, + "0.9": 0.98779296875 + }, + "dedup_dcs_abs": { + "0.5": 0.7234575491747819, + "0.6": 0.7351646627745746, + "0.7": 0.7393523969949456, + "0.8": 0.7415241060128512, + "0.9": 0.7448609655839391 + }, + "matched_frac_abs": { + "0.5": 0.77880859375, + "0.6": 0.7347005208333334, + "0.7": 0.666015625, + "0.8": 0.576171875, + "0.9": 0.41552734375 + } + }, + "seed0\u2194seed2": { + "dcs_signed": 0.752312421798706, + "dcs_abs": 0.7534169554710388, + "frac_sign_flipped": 0.0166015625, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.7494049668312073, + "0.6": 0.7518031597137451, + "0.7": 0.7523467540740967, + "0.8": 0.7522786259651184, + "0.9": 0.7523325681686401 + }, + "unique_frac_signed": { + "0.5": 0.985595703125, + "0.6": 0.9962565104166666, + "0.7": 0.9986165364583334, + "0.8": 0.9991048177083334, + "0.9": 0.9998372395833334 + }, + "dedup_dcs_signed": { + "0.5": 0.738610315209371, + "0.6": 0.7489887924166396, + "0.7": 0.7513059097691439, + "0.8": 0.7516051994607551, + "0.9": 0.7522101182063732 + }, + "matched_frac_signed": { + "0.5": 0.7941080729166666, + "0.6": 0.7447916666666666, + "0.7": 0.6730143229166666, + "0.8": 0.5791015625, + "0.9": 0.4178059895833333 + }, + "matched_cos_abs": { + "0.5": 0.7529932260513306, + "0.6": 0.7548549175262451, + "0.7": 0.7545865178108215, + "0.8": 0.7545409202575684, + "0.9": 0.7543349266052246 + }, + "unique_frac_abs": { + "0.5": 0.9610188802083334, + "0.6": 0.975341796875, + "0.7": 0.98046875, + "0.8": 0.9835611979166666, + "0.9": 0.9879557291666666 + }, + "dedup_dcs_abs": { + "0.5": 0.7236407069043101, + "0.6": 0.7362415516399778, + "0.7": 0.7398484998848289, + "0.8": 0.7421371714056779, + "0.9": 0.7452495124501486 + }, + "matched_frac_abs": { + "0.5": 0.77685546875, + "0.6": 0.7337239583333334, + "0.7": 0.666015625, + "0.8": 0.5734049479166666, + "0.9": 0.4143880208333333 + } + }, + "seed1\u2194seed2": { + "dcs_signed": 0.7549174427986145, + "dcs_abs": 0.7560818195343018, + "frac_sign_flipped": 0.014973958333333334, + "alive_source": "stats", + "frac_denominator": "d_sae", + "n_alive_a": 6144, + "n_alive_b": 6144, + "matched_cos_signed": { + "0.5": 0.7520714402198792, + "0.6": 0.754646360874176, + "0.7": 0.7550976276397705, + "0.8": 0.755016028881073, + "0.9": 0.7549161911010742 + }, + "unique_frac_signed": { + "0.5": 0.9854329427083334, + "0.6": 0.99609375, + "0.7": 0.9983723958333334, + "0.8": 0.999267578125, + "0.9": 0.9999186197916666 + }, + "dedup_dcs_signed": { + "0.5": 0.74111597246277, + "0.6": 0.7516985235270113, + "0.7": 0.7538686275947839, + "0.8": 0.7544630386255449, + "0.9": 0.7548547558641682 + }, + "matched_frac_signed": { + "0.5": 0.7975260416666666, + "0.6": 0.7490234375, + "0.7": 0.6806640625, + "0.8": 0.5849609375, + "0.9": 0.4290364583333333 + }, + "matched_cos_abs": { + "0.5": 0.7554351091384888, + "0.6": 0.7571477890014648, + "0.7": 0.7570046782493591, + "0.8": 0.7566967010498047, + "0.9": 0.7570130228996277 + }, + "unique_frac_abs": { + "0.5": 0.9601236979166666, + "0.6": 0.9754231770833334, + "0.7": 0.9803873697916666, + "0.8": 0.9842936197916666, + "0.9": 0.9879557291666666 + }, + "dedup_dcs_abs": { + "0.5": 0.7253111505221265, + "0.6": 0.7385395018694302, + "0.7": 0.7421578254288761, + "0.8": 0.7448117349607248, + "0.9": 0.7478953530274641 + }, + "matched_frac_abs": { + "0.5": 0.7805989583333334, + "0.6": 0.7369791666666666, + "0.7": 0.67138671875, + "0.8": 0.5787760416666666, + "0.9": 0.4246419270833333 + } + } + }, + "summary": { + "n_seeds": 3, + "n_pairs": 3, + "report_band": [ + 0.6, + 0.7, + 0.8, + 0.9 + ], + "n_dead_mean": 0, + "n_dead_std": 0.0, + "n_dense_mean": 73, + "n_dense_std": 1.7320508075688772, + "l0_mean": 64, + "l0_std": 0.0, + "recon_r2_mean": 0.9349369733975795, + "recon_r2_std": 0.013730528527824016, + "l1_per_token_mean": 77.01152250015718, + "l1_per_token_std": 0.1852703969732395, + "babel_mu1_signed_mean": 0.2886672814687093, + "babel_mu1_signed_std": 0.0009071832321226903, + "n_dup_signed_per_threshold": { + "0.5": 168.66666666666666, + "0.6": 41.333333333333336, + "0.7": 13.666666666666666, + "0.8": 6.666666666666667, + "0.9": 1.6666666666666667 + }, + "n_dup_signed_band": 15.833333333333332, + "n_dup_signed_band_std": 1.7017148213885114, + "n_dup_clusters_signed_per_threshold": { + "0.5": 79.66666666666667, + "0.6": 17.666666666666668, + "0.7": 4.333333333333333, + "0.8": 1.6666666666666667, + "0.9": 0.6666666666666666 + }, + "n_dup_clusters_signed_band": 6.083333333333334, + "n_dup_clusters_signed_band_std": 1.2332207155790618, + "largest_dup_signed_per_threshold": { + "0.5": 8.0, + "0.6": 7.666666666666667, + "0.7": 6.333333333333333, + "0.8": 5.0, + "0.9": 2.0 + }, + "largest_dup_signed_band": 5.25, + "largest_dup_signed_band_std": 0.8660254037844386, + "dcs_signed_mean": 0.7531264821688334, + "dcs_signed_std": 0.0015531529836029165, + "dedup_dcs_signed_per_threshold": { + "0.5": 0.7393597284907526, + "0.6": 0.7497618144455677, + "0.7": 0.7521397245646867, + "0.8": 0.7525687698054955, + "0.9": 0.7530362565173871 + }, + "dedup_dcs_signed_band": 0.7518766413332842, + "dedup_dcs_signed_band_std": 0.001599140639249882, + "matched_cos_signed_per_threshold": { + "0.5": 0.7502273519833883, + "0.6": 0.7526611089706421, + "0.7": 0.7532841364542643, + "0.8": 0.7531816363334656, + "0.9": 0.7531587680180868 + }, + "matched_cos_signed_band": 0.7530714124441147, + "matched_cos_signed_band_std": 0.0016006721232204198, + "unique_frac_signed_per_threshold": { + "0.5": 0.9855143229166666, + "0.6": 0.9961480034722222, + "0.7": 0.9984809027777778, + "0.8": 0.9991861979166666, + "0.9": 0.9998372395833334 + }, + "unique_frac_signed_band": 0.9984130859375, + "unique_frac_signed_band_std": 4.069010416674068e-05, + "matched_frac_signed_per_threshold": { + "0.5": 0.7957356770833334, + "0.6": 0.74658203125, + "0.7": 0.6761067708333334, + "0.8": 0.5820855034722222, + "0.9": 0.4219835069444444 + }, + "matched_frac_signed_band": 0.606689453125, + "matched_frac_signed_band_std": 0.003772567378417763, + "babel_mu1_abs_mean": 0.33069799343744916, + "babel_mu1_abs_std": 0.0008778302224156741, + "n_dup_abs_per_threshold": { + "0.5": 473.0, + "0.6": 297.3333333333333, + "0.7": 236.0, + "0.8": 196.66666666666666, + "0.9": 148.33333333333334 + }, + "n_dup_abs_band": 219.58333333333331, + "n_dup_abs_band_std": 5.101061980934297, + "n_dup_clusters_abs_per_threshold": { + "0.5": 231.0, + "0.6": 145.33333333333334, + "0.7": 115.33333333333333, + "0.8": 96.66666666666667, + "0.9": 74.0 + }, + "n_dup_clusters_abs_band": 107.83333333333334, + "n_dup_clusters_abs_band_std": 2.504163200219453, + "largest_dup_abs_per_threshold": { + "0.5": 8.333333333333334, + "0.6": 8.0, + "0.7": 6.666666666666667, + "0.8": 5.0, + "0.9": 2.3333333333333335 + }, + "largest_dup_abs_band": 5.5, + "largest_dup_abs_band_std": 0.6614378277661477, + "dcs_abs_mean": 0.7541990280151367, + "dcs_abs_std": 0.0016383106590241684, + "dedup_dcs_abs_per_threshold": { + "0.5": 0.7241364688670728, + "0.6": 0.7366485720946608, + "0.7": 0.7404529074362168, + "0.8": 0.7428243374597514, + "0.9": 0.746001943687184 + }, + "dedup_dcs_abs_band": 0.7414819401694532, + "dedup_dcs_abs_band_std": 0.0016504245582726193, + "matched_cos_abs_per_threshold": { + "0.5": 0.7538286844889323, + "0.6": 0.7553350528081259, + "0.7": 0.7552865147590637, + "0.8": 0.7551141381263733, + "0.9": 0.7551379402478536 + }, + "matched_cos_abs_band": 0.7552184114853542, + "matched_cos_abs_band_std": 0.0015311249277098766, + "unique_frac_abs_per_threshold": { + "0.5": 0.9606119791666666, + "0.6": 0.9752604166666666, + "0.7": 0.9803602430555555, + "0.8": 0.9837239583333334, + "0.9": 0.9879014756944443 + }, + "unique_frac_abs_band": 0.9818115234374999, + "unique_frac_abs_band_std": 0.00021434842433375526, + "matched_frac_abs_per_threshold": { + "0.5": 0.7787543402777778, + "0.6": 0.7351345486111112, + "0.7": 0.6678059895833334, + "0.8": 0.5761176215277777, + "0.9": 0.41818576388888884 + }, + "matched_frac_abs_band": 0.5993109809027778, + "matched_frac_abs_band_std": 0.0032066108461448273, + "alive_source": "stats", + "frac_denominator": "d_sae", + "d_sae": 6144 + } +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/seed0/config.json b/lm/pythia-160m_layer8/topk/x8/l0_64/seed0/config.json new file mode 100644 index 0000000000000000000000000000000000000000..1ad25b8fbad399dcd95ee6f7012618be6449b4e3 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/seed0/config.json @@ -0,0 +1,12 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "topk", + "activation_name": "relu", + "device": "cpu", + "seed": 0, + "top_k": 64, + "top_k_inference": 0, + "class_type": "TopKSAE", + "module_name": "consdl.models.topk" +} diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..8e8a4cd3a5820f3881171eb1592dfb273bf16c1f --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 64.0, + "l0_per_token_0.05": 64.0, + "l0_per_token_0.1": 63.99985297851563, + "l0_per_token_0.5": 54.50724526367188, + "l0_per_token_1.0": 20.63741533203125, + "l1_per_token": 77.13320429374109, + "r2_score": 0.9271624524349621, + "l2_loss": 0.048493126000594876 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/seed0/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/seed1/config.json b/lm/pythia-160m_layer8/topk/x8/l0_64/seed1/config.json new file mode 100644 index 0000000000000000000000000000000000000000..b86d7e5789f4f17317558623f7569acf12145128 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/seed1/config.json @@ -0,0 +1,12 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "topk", + "activation_name": "relu", + "device": "cpu", + "seed": 1, + "top_k": 64, + "top_k_inference": 0, + "class_type": "TopKSAE", + "module_name": "consdl.models.topk" +} diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..d41f05488da7fc32c6d56af11361285f226b59f9 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 64.0, + "l0_per_token_0.05": 64.0, + "l0_per_token_0.1": 63.99988251953125, + "l0_per_token_0.5": 54.61430249023437, + "l0_per_token_1.0": 20.685556494140624, + "l1_per_token": 77.10306360961033, + "r2_score": 0.9268578212374675, + "l2_loss": 0.048695940613904365 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/seed1/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/seed2/config.json b/lm/pythia-160m_layer8/topk/x8/l0_64/seed2/config.json new file mode 100644 index 0000000000000000000000000000000000000000..ff904d2425408ced117183df3820ef7f96f15324 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/seed2/config.json @@ -0,0 +1,12 @@ +{ + "d_in": 768, + "d_sae_factor": 8, + "architecture": "topk", + "activation_name": "relu", + "device": "cpu", + "seed": 2, + "top_k": 64, + "top_k_inference": 0, + "class_type": "TopKSAE", + "module_name": "consdl.models.topk" +} diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json b/lm/pythia-160m_layer8/topk/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json new file mode 100644 index 0000000000000000000000000000000000000000..d92a7d5c1379c168e6adb7556c23e25245225d89 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/l0_per_token.json @@ -0,0 +1,14 @@ +{ + "n_samples": 20000, + "n_positions": 20480000, + "l0_per_token_0": 64.0, + "l0_per_token_0.001": 64.0, + "l0_per_token_0.01": 64.0, + "l0_per_token_0.05": 63.999992822265625, + "l0_per_token_0.1": 63.99781103515625, + "l0_per_token_0.5": 54.32708857421875, + "l0_per_token_1.0": 20.587459716796875, + "l1_per_token": 76.79829959712016, + "r2_score": 0.9507906465203091, + "l2_loss": 0.04844786001747125 +} \ No newline at end of file diff --git a/lm/pythia-160m_layer8/topk/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json b/lm/pythia-160m_layer8/topk/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..cf48ecb23b5a78a1e272572fe8253b3a79bbf5d1 --- /dev/null +++ b/lm/pythia-160m_layer8/topk/x8/l0_64/seed2/stats.fineweb.split_train.seed_no/meta.json @@ -0,0 +1,26 @@ +{ + "dataset": "fineweb", + "split": "train", + "hf_dataset": "HuggingFaceFW/fineweb", + "hf_dataset_config": null, + "hf_dataset_text_field": "text", + "context_length": 1024, + "num_samples_requested": 20000, + "num_samples_actual": 20000, + "batch_seqs": 32, + "stats_top_k": 100, + "stream_seed": 0, + "sparsity_thresholds": [ + 0.001, + 0.01, + 0.05, + 0.1, + 0.5, + 1.0 + ], + "model_hf_name": "EleutherAI/pythia-160m-deduped", + "hook_target_name": "gpt_neox.layers.8", + "tokenizer_hf_name": "EleutherAI/pythia-160m-deduped", + "d_sae": 6144, + "d_in": 768 +} \ No newline at end of file diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4f9c354013e572e5086226aa161af18dccb7b3ae --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d052a590d0b8ae55ccf73f5e715cd03b258bb981c1b475ae05704f643066c08 +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..eb53a063046b6583d4885379b895c8667a5468b1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:867e9d17f29067efca54281cfb31783672648d9733cdb18e39221108a493a565 +size 12416 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0d35854cc0da0f7ee04391b4679c1ad4af471d31 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bedf125deb41b3e617910a17cbe06941d1eaa7f08c462284cce9798c8bfe2f1d +size 6878 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..511a4c4d1e0a425f28ae7633a2bd521d7a3999c3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53970061a8974a711ef88a81043be8e4a6cbbdd6e0fd41bdec595400a65748ec +size 50471 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a337be7e5694741a94f4b459a9fd7e4b1ca36005 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fedc3163158e484251f6b4899c4a7c9851caf4f93a9ef6542adfa5910deac388 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..6033a1742b391871d9a6a4895977e770d1100a49 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:132b583ea265d70a538e50a7f9336235e37aa8a5b180f9d9dee9aad9fb205496 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..46dbafd1dbc64038237f37830ffe9dea6e8ac009 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:785845dbc1633abc15f4e23fb0285e3a1d4417604368638e3bbee0a8a94ef8cb +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..0201e6d879e6d7e6d510056dde2772a1f6381989 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95b911dbea65ffd24f5dbd9d6c0a5420b54f11b847dc63304e4b804bdd6da29c +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e2ea46f9f53e9831e814531a927fe685a0214253 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7e7dd46070a691ed03a37ff71df36f485e45c42e5c469ba04227cc8705d580f2 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6beb7daa4f29af7c593cb9c7b838dfa9cc130e6d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ddf34f4e195738a73af55358beb8d5c27a32309c2f376b2c1d8728f882cce1fd +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..69f386d428e0f181980259bf29e8c433eba47881 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e6d1a5ce7fa3e17524e6574667a240414ee08f1bdd8f3a28d77e83dab68397e +size 11328 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f07f89dc471bec2ffb88b936069f4dedbe3d7bdd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19bb987cef48a77f1947cf4057a5022352f747727fbbe886d57191f3567f5131 +size 6366 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..a0a2add11e12ee02d87a9962ab45b49bf32ccd63 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cbbff4f99ed2034fbb25e594192123dfd4577e57cbf9ea024b2f8476ce0c526 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..03e0f43663d9e74e7dd4f567a0f7f7ffc96d7ff3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f76100f3b789cdfaed7823f5ad2277af9bbbbd07ea57aaad7f188bab02dcc55a +size 50382 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..49ffad1e695139cc4cb6428f816c98e3bea5708c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:057b5a0f78fa8a67de60fb2e41b47d47ecd8e7e148820279b74da0ab72b10386 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f40c0e6143a01acc9e2c8836920dbe66ef77f9cb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:337ba691185df9f0a6c0c3acf709c35e1f23eb3846fa629eece2e267faa511cd +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..e50655aaea8685fb5f7049c3a18615d1613efbfe --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98c80142af035b4d0c275d3baae50814bbccd4636b6d20f3b6827f026e9a7004 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..85cdde2ccdd24012e23d2394e56aa3173c9fa0e8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f4f20fb218f28d613cfd49ae82a9947069f6670586976a0dcd40e255d9e7ec0 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..75941d7e32a98619caa61b3904695d3242f44622 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ad06c2fc4bf1dab7c57417ad08b7ecff480e3d8b0aa18421e6e9daa290d840c +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..75efce6a6d8e4e907310aa5db10fb38c3ba6255e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43017220cda005196e777890786f49895803172e2270fdf3f65131a9236516bc +size 11200 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f74439d2945bcef18e4d33f33672f0b940ac2a82 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9375e89e4c52b3f8cfee3f1eee07bdce724af967b3e311b3ac0eebd7bdcbf9c4 +size 6302 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..3b000578e4ae1afce2f47a51105959f4359cb921 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9d80c307fce57ad2cffe9f04fbd752344bb39694eef8731520a1e00063a5221 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0858a5d3786da2d154cdfc2724aeb6e07e990454 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52896c1e6c035bc63234587e2d8ffb2465b44bf97761ef2538259221ded67fb0 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..1a791279f2c49b44b48afb7d79f7214dce62bc87 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fab5cc22a6576a2ff0ebcb7b185666537188c33772fb0a25d3278971956c43a5 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..6872fdc06996c52a0b385e518097b1fa11e84649 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:47a8f368b76d4cec3af31ba4b53bbfd6ae7b2b78a300d3295aaaee9e74df1c03 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..2bdff2f4375e7f7e4a6cbed7a142c9cddbf14c2a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:885f1a7bed7c7f54df467dcebaa890f0a4e1860140282361a50ffd58499a89e9 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..89bd3686df71b5e1bba10f360f063033a2f08e11 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a7921630fabda1bb0de3177d44adc03608a18f6c55c2b38aa736cc9180eca1e +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3369451cb047c6e32068f3da9d00c4c52a6c6b47 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:071a604f6e87b4e23505c33a34cd9350dc65fc3349480efc211c6286e94b3bb2 +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..11eb1f3c959c0116218e159ae592aeb64f71b4a9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:205114ed0ba2cb7a610d0852a604869435395b8372627ede2ef413079f2b9627 +size 6400 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4a6b9f121236f4902229a60943ee816c634fbe4a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:661186c6f2eb4d8daf7452f4af58ae592b3e83d4d5b0fc0a4a1fe8ebf4e2b8ae +size 3934 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..0d90a74b1156a46cd160b61ed38def2fcd443e32 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f7b81e81461673365a37e6fe8b9b6c7c4528af3f2e782213b57eee47f70b151 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..612e0ce99e54b2db3d676c4ed66709a565be08eb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e507b73fdba5b23a30478f1be2072c720b9e3b9d329a29bf514f1447b09ec05 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..faf9c8f14cd0082d014700c6fc45e4a0f27fb8a4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:379584dd3b9f18ab325f8beef17c3153c376c79217fd4dac5964ef3c036d3449 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..38b0f41da583af6a0a3883b61516b80c235ca5eb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efd0ea5337f9d714310a07c4eebaa09b036166dbc033551f539e98d2cfa40f0e +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..06ae8f37007792de1d28dd515e2b10fb9b8e60aa --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c501ead83c4ba1ebb8b7d8f342729a5e9b00ff869b7f25547de39a54fe0b4496 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..84e732b291db5d7a3e9dde8f760e60fdb30b2782 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c7844a120e7622cd34fca4f6d81e2260d6a0b9f821301bd8a899847acae52dd +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9482708c6697b22bc70182322b7d7aa62b28d7aa --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:705fe6d6ae2079ec9773b30a6215920dca8c5cd8c3b02ec0dc1cef4cde4bb2fa +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..466877eefd57d3ef7fe56a72dca202563bac9363 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7bdca012e6d83f328477f02eb4b6c9d22b0a082a8b60cd9b54f167be03255db3 +size 6464 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..eb40cbd00eb78967daaf09bb2d5ff9570ccc63f6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86cecd2661231a2070749c9042ce945b986d3599683f70610ae8149c6d57875a +size 3934 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..343740e227b948dfbc563436628d64f8654b28aa --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2296c3b319fcf2128c17d8167bfc9b860028cc3c159f48f7c7335dcf9e2c070 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f353618652376a244a79a2ce88dadaa291b81671 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:755adb126c0fccd6742535b04c159b1b3d4d86355cc80330cd9e20d69d411af9 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..cc67536c2bc714556b2aa4dff59be5c26b7f6c8c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44566f85e18d353e2600613bff0ae6cfb164b6f1960c34c5e66bef4c3cf580d9 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e154c905e501c8f24cea94d3019340eb5a2a5a9b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0553162de3502d280ade6cc35014f4dec9e8f069cadddf6315eb910f3d57c690 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..f0c2405f1345be162e959cfc8f964491e7565a44 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:773b113c713bca28a690d0141238fc2d4144c00a27814b9eb5e55cf8b2fd4bc3 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8f0a2979d5fd95cc31553636b0ca763d45924b49 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45f101a8e1342c74988d4119a5c9f264c595547f6d847e7ca7f78ff4248209c3 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c83f97f12a9e3559e0c5a7193cd362c577c8f36e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0351afcf29b503d0da88bf23b7f3efa98dff6ab514fab000cff6158d9d05e443 +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d38b5c5c5dc2428f756ef408349e055bc3d190e6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fedaa8895d28bda9a6a933c7074582be00544186645e7808979a7876a90cf7a7 +size 6080 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..78b7cd7a078f4a0c266a7a10dcd83412cc16bc11 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:663e66c38ec88d974803833cc4dc976e24013813e4eee0335b8cdf1e963dbf45 +size 3742 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..f2194d23e4cea3232ecb05302389fed89d72a6ea --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20f80d5bb59a4f2e34edb9caf50fc3f8e55d6d3594c70b2c8c243005d6d19437 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3de374c3023efe462d78217eceedd081baa2bd69 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a3062cc0fdbc036ad811af5e8a3320bb77e637271c06f6a53d20f78d643697e9 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..9d787cb3a8cac9c0f8e44201dfd7ca1b14cc8f67 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a44c77c4c93fced67a5688e011006a6579b70276dc1c00722b03271dc43b395f +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..cea919b495ced40e0e63985a4fe4a5641c781852 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7340e143a2a7baf38c878c6a51ae011306343c3c1f179c23f0375b11d016bdf +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..ed08362eb4f8b3dd33ea4036045d9625aedd7062 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4511da421d0a91e3d79056e6ef5f7389addd4a26d97d9f057c7f8578bd0ac5bf +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..814d943c0100c8e5592b28e9e48ade49b9e4b52c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91880fc94780fd5072c2bee4d7339831a33dae36c448882c9925db2606161857 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..46a64e8e35a4ab8afa621bb0afa063e64b6410d0 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe58ddd613e844cd6f97ad5f5e9380e24458d18e797bf890ff69259aa7f4a369 +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..951f437daa96f6777e57b9fc723a642722c5eaa8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:702738abcfba574cbb0266d718327affc66ec12b5db98a465c137459c3a63622 +size 9664 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..1f09967d9e9e76411e1c798820e9b4f22dcf1f10 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9fe2ff37077848b5367a3df1c9a641722c7b355ee1ddca90419fd2d60c68750 +size 5534 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..fcc2ff47e6bbef7a2b1a8d45ad262b2b6290d24a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44419fcbbda6d048bc83dc907967d3315b0440923b18856a48ec863426aff980 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..cd8d83fefb1fee9f7be3f8ee3e93032dda429515 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3bfdd37701446b6ec0fb4fa5f4cda8a1ca9dc432ea4b74703556a6bb40626ca1 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..053452f9e6b2df1d6e79d87afd0a7694063522a3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3fb68e0fa421cbba27fa8581307055eb2b6ea7a8c7fc3e7074f75653a6a687ba +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5112e544bf2b8bbf8c81f43b58a29003a366276f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c470648e46f625ff749bcf45c884f5c7d27caa70e73118d6b9bb056446fb90e7 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..de9dbaa7ff0ce08f7f849d0f10406932c2f7201c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8453f1e5e58b641b51adc2b27058d47f43205f58b29d3d6c1bc836fee1a04c9 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..20e59306e0d7deda61747252019c2987ccf37148 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e073367a9a72279e4bd6e426e01ce8e1f569addd5dbfeeb72b6a60bc1dd1955 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ffd7b7f4a0aa743482efee655db5ae04842f47d1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:40b4810744e7151be5d482ebae73209873774b4f282fc23d54c5d885e6d876ed +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..61df2838d456681a82c9305a0eec0d5560a94d3d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:70d1a93e623f2d2dcbf9635a5352a5442dae49c3b32e6c341da76e976cd457dd +size 10752 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..44e624d78d6251538689ee1dd92a86b5460c159a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f53e055aa3ea4a90c99cb0475b890c072d7dcd9df5a24264bf32989d5f4fd699 +size 6046 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..696ea9f3eced8a81edd6b2aeee10dc5be6eefd2b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:251e170a11f1879a3eb19d21115402ebf9c386520e2b6f61eccd9970b19d01ee +size 50471 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a04458b1bbf5cf5920100d1220507d2f301500e4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6ac389ae8bfd655ef36ec2acd51780392dd6c1a6e1419cc9da014bd141d02b7 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..3909caacfb6ef6d2eee59917edf7ec860e61773d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:487019e0a29c71986a1235a7795a3352d214e92c54c82625fea91a95a2f8dcbb +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..71e1d0f628e73d8123400207bcabd5d2061e1ae9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d1df7949c20af5a4ec86b099013824298a7e0e6cca34eb5039d274c94ce06c9 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..df42763a96c6d45fbad85aa8b06a14abb34d2cdd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d7f3f1fd0b761a02ec53631e2f37351f5ea9c5456e2e143c1a7f40d706d7780 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bbd3efd48c97968e1d678a5465aad9b583dbb7bd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:943a7ce327bf3024570e4a9871d0f2b62927878332e4ded0b87f42504a4a7f67 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..db63613106aaea361c2ecc32b1f7794d5df94581 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64458da899dfb79b864feffd2631d985c4277d22e8e9260dac1cf53438c319eb +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c4debb5aa5fcd66d4119393d3288874f476ca81e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4265b0f5201216604bb415d81ad93cfad1043a372ca9a7e0b4cbd4c6be2a3683 +size 9536 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d94475df7084567ac5780e19a1e6e0efd043bc71 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7cd389fff41f8ded7d084e1de96abf7171dbe7ce46a849dee0d1a9ebbbef894c +size 5470 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..487f0d80d9f213e16b35ed026e169dd3594674de --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dad9f1aec5ce7d398dd6cea2aab84d49765dfc43933776a1be89f9957fccece1 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0eae46a0e4770bf6a043a7b84a7c0ffcb6627f9d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2abfa8fce33ad251f1ddc435057b5a58429630745a61bf97d1711198530ee28 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..69fdaf0878586d2ff721d31d0111642588c9a243 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7037db7deb649735e09b36dd3690bb579876c1e91ab93613f7c7b72e3cfe65a4 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f4045b127e3a4a790e055485a836b20867e8b264 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9905839e1ed1e8b9d187d4f5b80a8cbe4caab2b8fd3d6cffe590276d706e6f13 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..ec542326048be4a2b4287a7c1bb541b99a99f504 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a723d0a0a40ba00949a767d381ca0a7255aee6fe5b8f56b3d2fd9adb286a9a4f +size 50337 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3c23b518af5fbba51be2da64984d35cd44eccad8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9682b4c95db08653f2899dd659695f2cc2873e75d4434e7ccfd79dcc5a3a75f5 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..109ed47706d31c2db7416be8fa17f7c3c94610f6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd0a9b3f0f68e11d4cbcf80c448a6a07104dc18950525978bde575e52b6e262d +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e975b8238d307758ae1d7653a89134211a2a77de --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4033e690184900228ad44d361dadd2db7acedfaf189602be54789378b88dfd6d +size 6080 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d92999b3f4b1a1e867df8b36cb003c21daaef6d2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a08beaf5714d2111beec3a63ad03519821a292da0c61147831472140a6090176 +size 3742 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..306725d17b8862a2a204084d3ee7716967f326cd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:098da58bdc13c5df76ba386c49dc041ff29276a2b018fbe5560860f82fd35d28 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..550bdb48065bc0bea5446703062d8a413a250b3a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f233d03c13f0720b81ca1ba098998faaa20cbdc72583e6de5546fc6a8f40706d +size 13518 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..cad468549820218e35d65d61ebc1f568595aa573 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2f7bf66d7cc4aefdcd070370900db3fa14f8174bd82e2f2b6079196f3480724 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a0eb07bf5966566f71efc731880aabf031a97cde --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af8aa6f23acf1495791eb9804732fd81951df3c2dc988d909a9523ba164fe26c +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..93b3f6dbdfa8298ba61c3e1852926c8eb3d8c9c2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc5aedd39195393e42e98a731ea1252d53eaf7a22c726a1fcceb0e25716a2304 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ff3478de67da21adaa0432196cc7ced1b875ef9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c641b625634bc1b20d91aaf39da15639571236e8addaa6110464a461985ce7d2 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f113682a4e2c1ac05a3c558d99c0ad961959317a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a27a51542f0c26592ff41f4bc0b993d1192c04cc4d1fc1d382b21a55f90a01b2 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d7af314c6a58c666679b131368bac313a2447c3e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:945b044574c507012a03a606eb55d9c86cbe24bd392a4bd48f0e912d2b22b5e1 +size 6016 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..37c9e5bf5e59ada2b5c5f00df225211b25dcbfda --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65c924e8fe872810149f76fe7e9aac1fe91cb6ab26a685e63a40f39a1a37350a +size 3678 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..510e3368a72eaad80d863b3210e5d5b114f111a3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e3588f61e9393ca87ba6bd94041bce49b05800884ed05a410bdfe25fc907340 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..837dd0cec234bc2a64cf21c050bc7a3bf1bddaac --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41091b3b646322c545fba4cb2f04c0a42827d946a631d7b8631adbd7bf02a543 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f67069696097f040aa14a75ac0163e7bc0190c8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ac5f93ebfad931d07a4f1caf32ef716c3c513b098d82197559d1dae86856a8b +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5acf9d3202826b864e49b3eacea2de64d296c306 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d75c2b60d8274e43029ce85849d2355d70852e39362566a1d846b4d4206ae07e +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..021297d703564f266223c947b10e69f49bd70a58 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:13e365466b5aa88e8dd3b2aa153708d99b1d08eb9dc01d850758490d566c9fcb +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..cbf8ef7745c93f89256137e0c139bbc223ae5d0e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e8fc952b43dc8cf55a841ac531412f8b2b7b70f27653cd5123d8e63f4579e2d +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..253cabd0723cf75ba06ea9eda8620f0bd1fd00b2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:424adef7f7ba171907169b1f0d37361950604fd638eb0853c8d8bcfac3b50818 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2438f25076cdd172a0646b61795a3712412eb6b1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6aef0c40d0d840be4a7b62ba60385706b5b745ff2241dae0dff7681bdf546fe1 +size 6208 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..53d591dbbb65ec45660b7459a4e4586331501d7d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9127d12244d8e0469443772ab7acc287b3ccd5d90eb4a3b9ced80662cbbfe9aa +size 3806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..90bc1c1c58ea76c67aaa62c8e9b32008e35320ee --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7161ba468b8dbe912b5f7d8db8874554be65929b77912b39b17f69d6b5140e23 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c05bb6e40a692f51e005e0f505074c7104e72dcd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:31aaf442e9a90ad999601bd6526cb3c31dc99ecb3d1e522ad23186d45f4cadfe +size 13518 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..1822b1bc193cd5fe5c1c4ddb2f3c92ad667e2eda --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f635ef41858a21a9dd0d8f498088949d51bf043737d61f5fbc203d4ff4f3427 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bcea7dd9ad4614885d9cd52434c5d44b8700e0c5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b81819e90b180e7e54c3c04f70d92e94c7ffed121f8f98d180f109285c6b211c +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..9ba6d4da02808546adf230f9a2e752ee7087ff51 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d87a04fd61ab46fafeb5afbb61dc4666996269d1bf04f3fafccbd7d556ed9f09 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5e45fc528ee0360358f0bb25de887f115ccd0e45 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84bb2f3732a8f8a1a1c4800c207bfd3e7cf8d0cb65a42fe5bc7e00a61a06609f +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..915f929eaaa50a7d96eafcaa8e1d8fa2ef019ebf --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3d65c359177da804c25d1579fb2e33508bc5719278979b84887ba513326ccc3 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ab278b788a9caa0a8637ede6c3579f34adc2a54f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a0463eef1f131924e07ebd8a1687d6a7e9aea44a38ac24ab0af4bf0371523d1 +size 4544 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c62999a926dd3fce0725c17f0bed1cfd256377a1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c25ab387b527ce44d03874cccb78ccc3bb9cbf643d42ab94275819142ec08823 +size 2974 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..3063c6ea9f29b0d4fbbfec1c9e7606b1ce7537a0 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:599a4df193cdd71f2111a3270d420d3222659b93863952d5899f676f25de8cda +size 13607 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3d8212f2ccaed0229818db59d866dcb3127c7d07 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4da90ee9e47a702e11ef1a9eb921d323d43b5568fd8fe45830f25db3e20e553b +size 13518 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..52a1b6605291c352b454bc90552052b4754baa26 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c40197c4b491ecb69d88912ddad8540cb5214c6d727e3d52f7bd531412e576e +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4d91b084a47f9d71f5100bca5e54a573208cdbe8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad9187ddf36a9807e9f1d476562bf9877ffe2f1668e0ad06cf670b999e2c17cd +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..5a44a67599130e2d146917b76cd28edc75737223 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:49cf433097feb05f48e020cdfd47a7f0c5b1102b81623c65c6797734ed830d84 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4d35a2550be80461f29c887849172c8dff7ac0de --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d85194cda6d183e33787d0baebd6852a9f81f31e8d41c9e4b34a84904de25034 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..06a552dac8413c30dd4dc9bc7b3b9837181cbb20 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74ece9ae569fd445b9bdf8b0f083c42639c7cf4f8db4593c6caf2f51f65aa6cc +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..9f79264b5f1dc423d17050f7e22a0d777a58b6c1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:07d4b37a0d822358b19e45eb9e9739351ea2d545e5ce526e9a5008fdacf9a324 +size 4480 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..550289c4c952f7b8cb08c32afe8b7ec24c5ebb97 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d926f9b9a41f46368e04066372ce1a02c349a5c1706a750bcb95b010d151693b +size 2910 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..f44d81549073cac63034ae44a13ed0921e087bf3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0cb122a8b65387f0185853d8e2c27e167604aba332f10a313e51955c61572a3 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..de1076abd384e8f838c13abc6fe3b7373627db9d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c5e3277e2b78c21fc586c9761082b3f5909d7e82d9ed38d249f11679dee2aeb9 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..5fdc7fb14a1db617097acdcb90fc0a24afb095ff --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08ed3b5e461ecb599b6f17f21dc47e2dbf8648c75914e1cb89e51056ad1b0800 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b08b53556f6639969a1238c7e4fdb5c8d50ef9c7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e41e860150a87259409b40a252eeba208de09cc4764bb130a20c7574782926e9 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..1ea877ccb501c9365d7bda2966f2df5b905e8e22 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5bc968a85cf7e9403ea6607998fac13f488cc9330c5b64c7d6d72d6db952c6c8 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..44951f465306bf97e275d4ea2b682024fd17be66 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91cf75816bfa5834b485fefe0c9c3927e15848772c8cb7480be38f8c4488d632 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6d34b7efe07a7a72e1f4a3602fa42a559d3aea41 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cebba8b1e36289791f6cee686d39ba71f94891c370635aecf4e69146a9b4e860 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2e034d8a128a5502b3cc5d8b9c8fdbb4386e7dbd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7dc1d26c8f5ec2f8f21edc6e4692a87dfa534c3cd4ff0d56ff1855be4dbe2c30 +size 4480 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f79d4979c00fa26540551adf0c7cf04586cfaf0f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e54d84433a4c112d693f98636c11a2101064f49f65a58b0b4716986238984570 +size 2974 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..82f81ac297fbe6c9cb5301c505a617b4e2d42549 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:986cfcc84abc87b5b0eb027cd061f0e1f6983fbe9005d775b501a2c6263b9f44 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..17f1fe494d134aa95b00055f4894dd5b65f89e17 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:706e36a886cbdcdfc107376d20757d92b4d877981a4103352cf03bcbd6886c27 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..16556ee4be214e38835351abffbfed5c14949af7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c677f23a77b3717b498946c2a0fe70bfbe734b7b04760b717dd1bc6fd18f4ffa +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..da6841068bb22e4dcbb9d48e30890775db185e4d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:781690250aefa8a2c4436523c80944f439e1f6ccc321048e9811580df125ec9b +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..a6c1d7f165e6d6ad80198545aa280cc13753b3f0 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97f4fec441587f49162a6fd9cb335eb453ff7949bf56f6174f1833ac64c18f90 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..9ceb20d3edb397d16bad38d209d5c7184f361fd9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a82e46a899437738a4577a83d0dcac049ec9fd9aee365e1194fd8aa25041e857 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..03cfc35a4ac8444ad1db43651889bcb0d5112e70 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee2966b3270ed5286e32fde68b78c677fb3ac90778d9cf4d5cd4f30082537354 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..fbb80bac42276e7fde4f82740aec4b61677b1c43 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:379428f3f6481603cc7cb7702623980dd49030231aadfc3438d9850640b2e9d4 +size 5568 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3ea5c1a9e8bf998f9345bb0399f511a9454a7de3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:302b58e283cb3f76a6e5f3827d9818a5ef43f5509f22551b6a348953cf0e9402 +size 3486 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..7ebf35274cb08802f58fefe413fbd39fbbca5020 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4031ef3caf714948743e7770a33f34ac21016300f8f03b402a1913b02c1f1a91 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f4e753a7e658ac25ca32218e9c0ae064b903e441 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ed7d35d6603839247d1578639a782945b7241e9b273f514b6891e22a2e17ea1 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..d47ce3a8370d52a3f271881c8c86b0de571ca2ff --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a49b923b98d37026251229edf5d6c1a216abda87f7a51913a66e79707d6c37bf +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b6d58e89b882fb3a9e0c5b272c207d8b56536dc6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19d3c759a9991b78f3aec06d2aefa82d57d60999ff033abfcb4e4c8fe13092e3 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..216728f2f3a455e86648362d0c3719f5a8656cfa --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dbb8ed2fde54aaea1df570e1bbf3637ae069b809fc2d8ce64f5a8633709a9742 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0420e712603ed6ff548722ae9d73eb99ac8229ce --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a87e18f133accfcc1a2ba6cdf2d239e5cdc726776a39b66850304cbc6b44e1c3 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2760f0cf484e335d137308b8b9e9f9fe49668859 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:abf09daf06e97b2a98ad266027b8e0dd483ff48b971bab87defed5268b58b50f +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..dbdf401f9c32c9eed1fb45f84f0636386bb6e07c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43ec28c53f372b6db79cee25b06bb1fff0b5e6405b6125f10d3d9cbbbed3f3d9 +size 5504 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8c82092c9df041e97ae2adb4fe7b050e1f9a762f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b97cb2372a6c2ed68dfff9c2fa95987f4c61b2ae981066c486364f8dab630e55 +size 3422 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..2b52fbbf41db23eea59c821f802dba43d45110fb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03e2588c9d90b89c794315fb941df5ae0a377836bd2b68a655f0c5edf6d489e0 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3d7b4b9984b9af921542852a112287ad583bbc15 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec97606487aa22b92c78f355f3cf9cc478a0dcf4fc60fd4233b44cef35761a11 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..ef9db54b0caac0b65c29c0b1070411778745a35b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1437020c3da1c5061332ace965e02edcae1580749fc014e4ec4d94f5eae91e36 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..1e0fb2ae71fd148ea4b279c37fb8183c34718a17 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9651c8e4bac84a3844a7313652f07537813f50af912f157395ef419e5b35b3e2 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..b0f4f49526044d2c1fc5489d7b2aa776b78df96e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efa50e1d46241174db0ea5a6dff01450fdc7727024be8ef8086b21ab352f241e +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5aa709a29d465fb88bb110df7035b8c70a7c840e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7d888b0274be19bef212890db9e275700e416fec7a01fd81e149672dca27d8a +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..63879a2cbbb50a903d193a7e28c914c519e0ddf8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bdf1dc0c1c61bdda264e36eb2d72eb9fc7718b0746ebba8e49a54b7b4b4619e2 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d3e37b9486072845accbe1d2562676b34b8f6a6d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43d1f1a08078ca795188c92d9efa1125bcc896898144d6a7553ac4e0f36221c2 +size 5760 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a89a8d42bfdc745b237cf69cf0c733569b74fbc8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48f4b66c67e1f4864508d0511460f22a059990ade720d1efaaeae5945e519203 +size 3550 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..df090f6b0cfb5b30cd3190428675ee51531ca676 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7efa2df786f879b0e45e675c983171c253b5185fdcf1e83543777bf4640064e8 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ad247f54caadd263c50cff85394f10ba16931a17 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9eef0d59ad35193ca45f7780db5d25fff18e35bc0863a47e2983562dc51d52b1 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..9b0a626862862410a15b8623b0c7739bc6e8ae77 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba757ac7a08f49fb9b8b92b46e9eab55695e675ac5964477f0ba0fa55b593050 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ac3e0281c6498d568618762a465f4c100581ba61 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc0ab1abf7bb894b2e7a7026a76bf02413385ce2ce6f8d3c6fe8e74908af2586 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..bf002419e190286995b5ef31a9ae6c2af2c342b3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:392a17808c1fdfebfa4ed72cfa95ec288407ccd47f763516e58a58025a50ee41 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..05b86f56e491e0a2116fcece552dcfb73a5a9488 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c98e579914653cf2f29b432a458686f95bc8cddf44a7f4da3c7fcd3027da2ee6 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1ea72d0edf0e8bae3f3282265a7f4e7649d9c180 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a1f2a9ed1df4e396325c5300be47e9432c115fb3a41f4580c72d603af3c5bfe +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e97c7b32a1d823bf7da773788592e00dbca2b799 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:341e9dd612472ccd42fdc92f7a6d20ff8149bc4a22fa1d3f57002c102940e49c +size 8000 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8a1e944cc1af28be4e01bedd8d8775f69ee02abe --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a15afef845795caeb24e08eec57ef4bfd93c6574dee6eda92b93e142bbfaa1a +size 4702 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..3e7f402bf5bb9b75bac382e56cc909ed7c518973 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9624a3a1dee7fad75798bcd2cceab8f29d612e246f26d0e4db253c925e144b03 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..6a73d2fb171ce4c295d3062d8768a91c018679ee --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f31eff27aea9727e76e7442eec0f340c47ae112d20fa8cf57f9147eec278278d +size 25806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..b007d25a41ba26433f630b2e27f22258f89c793b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:971915eebc57b0ba5dbfbe63c35af3d718463e527fadae381a94139eb18711cd +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..042c8e3804d08f1571582b3dd1724c80de2cfaa8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92537f69b62084f4cf35bedffc271fe71d572f4458a82fc9696a4e88d5bec91c +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..b235c3ebe7b60e6e009e704ce646658a343d0c4a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:46744b6a353fdc2584864f1708cb899750b73bcc05ca5442715029902d5360a2 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..191b7a73f0c7555a9a18521ae1e371469b5bc5e3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:35344cc10f90175a16d4c6cb14c986ef7e541d362e989ad6288abe69155ee501 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..878fed8134387afa926f5a45aa4b9b42c72621a9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:044504531394f6512a468cca17f358d0434699ff0a55f738e404c85636696145 +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..eadf177f5ec7ef9b8c70d3d3628f91885b23c390 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90fc56019468aa7b6ed3617a4db86650c9ce68952b0ba20e82d90b57a93895c9 +size 7552 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..78519aac6dacc3ea4bc3e00f5ba5e863079d7439 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed1f68e1ca1d32af941232256eb64d518213b8fdf6e795a774d860c79781e658 +size 4510 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..b749bb5f044d9ed33410ae07d196563b8be2fa1f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93128e117dbf8e2bb73204dd8dbb1c5e844c1fc0ddc4175b230b17eb9ba16aab +size 25895 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..73964bff89b5f2095b87f2587825a487f63794d8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f506b154051479f0d4ed0e256dfa603adb5ad9349efe2bd1e6e0388c83795554 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..404133c1d6811696339f2190c40cfaed0b7548ad --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de9efbff6d8cb5216bf9a034cac8261d8e5da12f031bb69e552a3d102cd38eed +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..6744a528660814de524a28827182a120938bd4e6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9d310c85d63759e4c84ad7625e07f1602fdc6cab204aabfaf0c2ac7d943ffe04 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..1265d2ce3a0e96d4fe32d2f1e601d7ec1214af43 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eea15cb4d2e7681b3a2dd6a88a42497e1b0aebcdbe5bd2e459bfe64a3f7e0041 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d577f245ea3abffa2164519d245db07cebdd94e6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:73909d50607537ad7b797644564dbf9fc1921ff284edb96c7b95512880f9353b +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6ba447eae93e817f32f4c5abf947e6b56e669964 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4aa706fe99fe9bcb6b222785a93d57d732b5e523c8d04d7e097edbdbb96fd4f +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b053f02fde1e16a5191c823fbf4c9341a4a9658f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de377b044cdfaf36daf9f5e23bcd7da8fb25a0e10115f6494a6023d0aff39590 +size 7680 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f074b2d19c52d914e802688dfe91128767ab0a0b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d066e2583c0e4a39c0d927a5efe2d8551ca2e72837fb74329db28875dedfbe78 +size 4574 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..28e2ec5812e20a03075091ea7860e7e39116f757 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12a816e1593cd97114519a33e2157634eeea61e66f6acbcac934498b528b6e9f +size 25895 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..9829bb2b5656095ffd6f477c83222b78fadb88b3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:440290067f8150811ed827b5193c5558694c20994bb27944eb494ab9db4eceb9 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..ad51ac80040ac08a2a1909ce43319dd19bbf1ae5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e391356c372fceb1c1cf50f2001dbdbfd5eca2714c3aec218040c2b2af14cc6 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c2e823f1d62a9403740826bd3dc2b0f9308cecc4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12564bc14076ec4f90a23535c8ce001cae7c83ff6221cff4472d45dfd2bd7df5 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..73434beb52ee629351c1879e8ecfb8b8b92a3c07 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea6ea718b04b54e60778171b1fdac66161d3de4019c6c4a84011b2c292ceda27 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..89ebb16acb55ac543e52143b742edd712b1453ce --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8845f8d1269a83be3732cc4a0b0326eee16b529350ee38da76c01c2207dfc3b8 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0ec3b18480d3f91bdb0898b4573d251c737d5ad2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b65da9b6d09c4a7979808b39d90731cf1ee62c9ac0dd647b62f539cea9e7a9bc +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..48e42c5c404ad24eb4b54bcb4e8cdf4e2c799f73 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6052b25911f31f6a66a952c5d86be9dd041e9b9e2c96c0ecf7f92470ffa526e3 +size 5440 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c85bab2365eddbcc2021070f3f8fade6f91b26cd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:041d1aa83503c81777217a53bf5c22878f59141bcef5c0917c768be054bc269c +size 3422 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..574adb088d23c88adb40f611add02415944e361a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b5dcd46f171a5565eea1a6546956c17f6fc8cf4339db058c9e8d6eb5d61f00c +size 25895 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bd3d9476696cc5b288e5b28a691373e4c3b5ccd5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:732116214e395eff74972f49c1cdb82ab468fb697153ed5b214616dc7a7c66e3 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..ca9b763daa6ef7c86b98ce0f75376b8822f59217 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:30a1383b5bf92dfa0689c79b7d02f00cd04b74c72aa2dd6e5de0c09dba0f30fa +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e8c95875cd37043695334c60f07ffec9529c307b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f068540f1da45b4b2d033c43ed66bc4ad2e47ab68f80183dd2fd7110ec0d7ee +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..ad3d062f1386943bd8173399f86f16a96e419890 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ce9eff70b97641fc9aac125b4c8487423cb5073976eaf89396acbbb0a3e0141d +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..857dc956e12a8a5279dd7f9831d47a052c35a38b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:264eab4c44aa7a4f177084fd10e1b7700cf6572478db3eeb0135cec0a1312de4 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1e3dbfdc099d8a2ee70376f6bed9003c79678c29 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b4994ff74b3a1cefdf985f42fcab17d49861fd2dd9715dd46048ef8a79c6ee9 +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d76635a1711b122f1ca58117ea640cd80270598f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:370b815bea4421ed576e9d3210df40ca5e69d277f2fde445218326cd904864de +size 5184 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c540293f3c599b79d0aaece76bd424f3bbe91c9c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41c872b3dfa2b534426bc5f2ee25eac6f75e7ceabfc41f069a1d86b61560e294 +size 3294 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..2225e66d6a0c9b0b3ec0dad76730b5d5d01dd952 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e37e4671578d7ea9a9a61d1bf237b8be865989047235c109e58f6c10deeec0c2 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c308ec6432a38a042b8535d15fab208b4770a852 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:163352613c8ce9c84466dfe7713e6328bdf880b3e2b5ef79e469fef61df14327 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..82b369fb843fade48da82c1f6803f890baf9f752 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4ff88640be94f3a2817613de76c86aa7d711199d355eda5095bd497bf35bb927 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4aa7d9c89fbaaf0acbf1a6cc0e9de20aa5a78c02 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e2d9e100c452ecb41a7ec338354ad1728d583bd67a7a32d476e35afc9c348ae +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..c80f3a124d54e00200aae385a588dbbe1d242040 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1078d65353a30e6d84c221f19cedc93486d93fe580049e3d896ffd0dc8af5a7 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bbf0a7ff421fd13181badb27ca68d441ae37b862 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee762faa388707926d74871dc06193c8e3892352dd455b4ea66e619cd6d6c543 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e46d0478cbd7399f1d3a7568c8cb2de483d65d5a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7aa96cba64a39b271c6a6121d354b770e9c042f4d991260358c37b1927234e5 +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3b1f8de920a9547bb12670579a4a660722186a07 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bde8d027c6297c4b1e3d8ecb5585c8adda0066e01dff77f53170f8c1defbef53 +size 5504 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3771d910906e1cab240097c224232f3da5496b80 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f915cce02323b747e3fc0f979f41a2f42b79b2c8efbc387b3e781a2a11a14f56 +size 3486 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..b9dbdddc12bf07181930ccca0d4fdf8d2b3244c5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:40746fae1a26370a3b1161decac1ce5d9c1c094a7ec5e27de15a91ea681e3f12 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4ca3a80f9a22b76f019d2efcb437e936aaf70b5b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb49cdcb873cec47999d112a3fd214447d957c27d02cc1506120bd623e1721b7 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..60875acc97c3190a405778007f17a1954b30064d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92eb7c43b85b4a67faa3391fb70bea8dd998cba6c64ecc2eb019f3e54d4188a4 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2b71d02523ea21830431dd4fe6efaa3835eb7702 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e90a21874622738521b8e4d7013cd5cb83e41f615fb0c84a439dd512f9042098 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..debb16eebe134b81491a6d98a1e31db2195ad4b8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e43d2cb92f67825c08c3721a350204ce2ef4410714689c08179d1b83089608fa +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d67e8b9435b42f36aff24539c2d2af5394972f41 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20dcab3171ddf72b6fbe3107454fe1c078a1ec39769efb3b691876ab8d0aa36f +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ed5b6cf574311c585cc8ca01d81de06bd90420a2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ed874a25ae52a063ac9f1620dfe38043368f1a23c6e12ba1099dd9a2e528623 +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e283a50457a4c62d6eed399db5a916be612d52c5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38e3455fb5cca66ce9efde2fcaf667f98b12d3419dd575c50a5bde6566e53d6f +size 6784 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0a1360e7d3da5ed09b937b045cbdea249a8cdd6f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:350b6b86f12e2fdeee3105c0a57f22091acc19ff5983d2c741b99357268d538a +size 4126 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..6220a258640098ad08bf0beff341ed581cc4fcb4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a4b8c1d419c9b6ecdde8f631daee1935ff1b3e13d7caa311324255398c393a73 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..de60cec14c803b1b1fc4f02cd8891259e54a95d1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a94d7eeab1c67e8e0cfad100fd344420f9e404c64be7d7b06670f95e485e611f +size 25806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..3371f76200cca111e4326099c3286e3c0b6ff605 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9d0f3fce46ecfc24dd9a05f77de80fd3899f4a38f7509ba24436fe0a35b9c34c +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d977284500898e85d0e3dbcd983561bb4b094ca0 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9548d6579382ad415ae8c30a2575cdeb7e75ee64d7369410a42bf0eb3a76432c +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..975a58c94690364ba60ce06e1f51f4bdbbc23efa --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c535d5493fa64f5725e23c21f76ce3a9f370aa71533cee9ed691fc9f9d8575bd +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e8e80d872e9a7629fd0963fa9d2f5dda80089013 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f8437c8d3080487433a8fdef34d6308a81357369a942259133cfc9171b556ec +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c2d2b862382c8f9bee2cc1b7b4d8216d46d3f5ab --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a43b9bf413edc877062480f86071f943d7c26b41cdf301d2ae46eb29a016e14b +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a960b1bfa7898ca2710489e89681fc99ae590874 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec614d60ff1cce52f3854d6e29c390e7822870b2b2472bdc7e203b435a70fc1c +size 6528 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..482ebc5972e9bb739a54eae486220475bb947ddb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a67d8c4170e5b0fe8fd81c8bfa87a32f80d150eb109303774d2df05db754615 +size 3934 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..276d1b04a09a021fb778bfc653b69a34ea142d6e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db105070a7c9363d8e34d94a3d1c98464e0cb9b9c74599587b63bb4cc67c4d56 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ace788ed9f23dde35c613d8276930fe914d5d9a7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aaecfa6872095bf717949bc38235c55147066b8408e5c1a523a0cbe4e60303e0 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..0fac9e0a9430d942b1ea2de6aaeb05e7e617988b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08d6d711071c5c87036352acdd2e968e2ce9f3668116d044e3916aa0c216ecfb +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ff41d9f12b4749d0bfbb2d1721ceaa715b88f4d1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a6e15afd036b15df6b262f4ede2d9e9aa91a28afa552463623b4057d83a0c8b +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..a487090da354146f4ca194e7d220a1eb5836bf40 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a39050f909a624b58a7fa08b3ecb1ef4196944031ab20fe65c327fd24c54c377 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..78501791ce25fa586fcbff1ae4adde826981461d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b4baf8a8191191c4264b117d2093d62a8321453a6ca86fd8bf71d526ee154df +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f849ad56bcffe99f7f8ef435e7364914a36b1008 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:60156af54d2e30b0ae08bbd68de4c8c52de7e10ae678e11224108383e411ebbf +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..57dc8e2c493524ea42a6560c11eefef7b5795f68 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6cc65df947dbe284d884fc827c0011ee79f4c1973296ef6180b055b3c0415b00 +size 6912 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c1b5421401b458f5279a391f20a56d44d264f891 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71bf858a373b1207ce1a6ff230067f8779e64783182ed0c82daf20c378e2e304 +size 4190 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..f4f38bf69878035220f60b54d38d55057aa23b8f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aaf19fe54d4c69c07cc24f8824acb172f7447856e9d7470a20a0efd381d6c1ba +size 25895 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..39e163853385e888f5b01c2c50620a235b82313f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78493db1d23d1ba3dd4ce896aaec6dab1fb8eb6c5c6232e8381e377a352585ff +size 25806 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..b32196e780c9802e6da9e667d3c79bbd3cbd88d5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d974da379281b2c1b7cf02441469e38b42eaa3061560f4f7d93aabe550e638ef +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..cbc053d2257b0c1b9824a90bc155adce8d1442fe --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17a79e17fc255effcdfc158f1b55fb031fd749544e2b1bb9c1e6389e7fbc8aa4 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..7cdec65e9f9a48ac751c83e76212dcd9b12dae50 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df9ce126f97bb5122e1b6cbc7e1f220327728488a93e7dbcbb8be3d69f035339 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..11775d669d9f0317f1385d8fa4249ee391c75109 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/batchtopk/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3833d62b704c6761125011d2f2c9233783a77a64dca85a99cffa8f676a3403b1 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b7e07beaeed3bc84a201e1543b2c0eb6a42e6c80 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:977744bedaa60a1e159f4c95085de5d02a0517a05e5169355a200c1be6a9140e +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..495dcf188e9ca3845e9e533963ad1cf533505aa1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1cc6e7280dece2177af0bb7c800ef8d974cbb5838e62bea75d7bae07d03d6224 +size 7104 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d4f36789856c28522d4497fd7edabac519cdc3c1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:396dd5471a3716721b1facec7bead1d1196903058f22b092284999c8f360fd09 +size 4254 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..da3e670f5059b6b1ed27ee9bbfc0ee824f8ce097 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1aae31f557fcd377fcb8d8c0c9d19dca209c57689262422805aa14f3b84c7b35 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8e448b2ac1f2c783590b5d724a306aaedeff0e72 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e9c057ee5ffbcb54fc2117c63272634aad0e1ce17567aff3bddda3c53fb760b +size 50382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..d4d798c2c2f0f6eaf261d77466ed0323f80b55ad --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00764f9af4678e2fd00b25150f3f7037153122d52b219eea7f72770a0562f08e +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bc72f30a6c80416ab8303283adffec990e3618e5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0cb17008480f128ccec2609dab79c7259ed12afe2ebbe47f40dd098ee8e30d1 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..60b92003b2a8cfbc972f645a283f14fc0b3a1f63 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ff71d536a9c145df717b1a91471fa0877582fdbebd91292ed323d6f57abd7cc +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3ad200ad39d08113b8b83b3e2ee0a444777be28f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:280396e839a61f26a590fa56ef28f93af991eb6ceab9daff710ed0969ac8ed37 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ba41fda8f0cda3f09c317481d04dd13df928afd9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:35d51730f0d7228c57fa216d58d939998643ff82bf4e3c0151d2c52bf260d2af +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b9ebbbfe4eb26be1fc16864df2a3c2ed101d9a7c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:886a1951e1fd90939ef370876c4c6c58c813cf603284f35bb7c865fc8cb68e93 +size 7296 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..66f0e5c10ec11d90d980d1731b620d04a2c3a9d9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:39c6a40ff464de017c385eb9637ce33fce2549df8222c70eb047755840aa8e84 +size 4318 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..03b4ecb34fcc3f70fa60d59cb4af1331dc6592e8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72e104ceaa0155b73b9d59ed5dac6400274c995c1b5897f7a94ea8348013153c +size 50471 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e6c2faa6ef159186c5bff79fa802db893031f64c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a23d66ed8478bdcfff2ca05ad849d76efa5c9040c5d0ceb38077f44e4fc4e224 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..3cccad8b15665543b1cc3ab9f9879ba58b048868 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4af0e7c0d7c82eb2403efa9cc5d1d8bde45e0e35b0963460c297f98147c741d1 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3eb37efcbdb9983a3a1f556045576d95a4f0b0c6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d22938a7291e5197a3eba963cf3c4ead6f1430164138e006732bbdd1c8eb3c9c +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..fab966c8d60a16fafbfde340f60b1825158bbd23 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f548241c315321c431a0c57ead3355ee5c5a62eb430dc5f41a3028eceb10f7e +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7bf7dbbe326f4d27ecbc8bbc8bc2a82646ef5c31 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e3a2d9fe1392c33654bc70c55337b9fb99321500c39d530a0fc3311378d0911 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..edfa875705c9f0886616812a6b2af76a3080ace4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5d91e56084af3c6041c18bfeab0da4ce09ae2ccff887dbc30a4aad87ca817493 +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..741d549abd9b44486dacfdd2192715431480443a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e61ac4869b703188f7efbee83a651123d2ac302e9d8c25228f6f7869b58209c +size 7040 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d9ec8ecf561b7699496feea700c9e81716282823 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1af32f09760706f9d9ce204fd4d814b69892b6b63fe41dc4f70f9c699df40aeb +size 4190 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..234a4dfe20cd4e7c27f133c7e53c89ab83046ac7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:133bdaf072de38faea91c2c2aa5344297d723e635c426fa791bf3ce87cdcedcf +size 50471 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..32d47af6fbaa445e5d6d6ea272c250180e0d3683 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f07d336a686e4df914e40b21e8d09870e8f5b2841f112e35fd8c1d524cda276 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..90549b9c83a647f5891b92347e9ded9b6f415556 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2f1beb626fead379a41cdae6606311da724e2335aeed8cd1fa8ffa3a6f13935 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..cb5101b9fffc3f2a9d5bd5648e509dced3154b9d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b3d3acfcf61661a0a6bb9db5da211bd83cf9253fdf8f59532f5e9c26c0bdf78 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..09a0b27c1e1a01138095893b32702481bf0392f9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:505937e8648fc1597e3be48785c76b062a48a6a05a494e16bfb2c9d7ecdc14d6 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..41debba722d7fcfae704bbec344e8d236a1b90ac --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:70b161b64d32c237296e3b79b1a2d889121b0ad18315b90a7ac6c36694d3c749 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3ca39e445b526fb06d4af74089383a1362ff50db --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8f6e8bb2c43d089820fd45b14fed0c9db71a2e2e8986dfb68ad780cd6f85a3e +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5a95d5be0795a518e18c1e3c7312a4409d7ea39d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a9a1e369985a163d9eaca9e5da94966e03a0e78caeeb76d1680100b8b99c5bf +size 9792 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..90dfd9f851f9caee628b4a563705554490783c8f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fcf31ec9b264f4ab2ece20397cfa251f9410c567891515a391e55f914d668f53 +size 5598 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..946e7d84142de95ad47f218e96d8cc53195c1c08 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5334c757a9319f4c1a7bd74791752f85530e3d14e09a971adda9a764c8a97ff9 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5b168725c48845144c3e659bae3a4d68c80592c6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b54b8fe75e1e35eff8b089bbf451738747b14da3f921d78758ddba27f54b78e6 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..1f5c770324be7882b16b6a8fcd79be5f12a39c5e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f441cf90d961a2a3d6507af8dffca63f493cd9f29a0f2cbb633a72bf5f7a337 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..cd493561a8bb78c6dd6c790c1a7856368f737c06 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6496291908bfb67937f15148591c5803ca9b73caf3442180d4464ceea3847d36 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..6f2d2659ebefd744fc5287e83c48a66c4f875d9f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:62fd46a2a7dac9dfef96a4d3f663235963426bef4c25328181cf6e2da0f85518 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d4a6d929f9edc36cf15614154117e423231ecbdf --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed0b0d917e76942a5db71251c9e4bf39f1cedb299089003d299c3e93f9c63ddd +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ee64d121b4447356e82a56e561267161f134b100 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b59adc7ef844c992b3f149f4bcfc30099441debd15bae06d29c730880534511c +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..9c73a8e299afffda28ecaae544e20793fa4d6c87 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88621fa85113af7a504d033444022591a0d5c3011cd422807b7c0f4abca93083 +size 9920 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..52a7bfbf07a78515941eccd33bc5d43d1c3cf87a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:925b7727c41349a022b5efc269d34a2fb54a20ec92877093799b17b0abffb514 +size 5662 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..1162f032ce2420107b200b0e7ca217b63c7d99d5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e43551e350e1ab611a10c691ced35a7d2198bddcc5abcb98db86b8bff391e25f +size 50471 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..16f68ed2031b26305bfc2550c7956c2c1c973fd5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d538f3dba07e43266fca61305450dc4464129d5f577917886dbbeff6c7243db +size 50382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..8ec42f27e88f8b408ae94b96f5739380b758fb43 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f96c3eb28ede007de6ea0238b42a1e29a2a696cde60acfc1e75d45eeaa9f398 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..89572b73d57dd3b65f1eeee66da457f94b387164 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6312bfbf05f76902dc194868386c600b7e33fbddfe64b2bdab8092386d158877 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..15ff3b8cda67cf7d8e8cf6ca395f9816c9977583 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:847307a2646ac149c1498e627f8fc4a59f301e93c3c70537e4d79bfe0457bc4a +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a4c45be80b1df53a298698c8781b32665b623879 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:abb41e10879015653b71246a7622796bdb51ff4437455156d2cf3aa1f6c1e28b +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8c7fe1bbfaa46a21f4c90fdeb23252ec7dfec52e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f036aac3c1c1279ba71591abcabe73eac887b585f6474dd19d097f36a514c35e +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5cc5aaf867275d0fc11e2964dd48a79a1cd01f20 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be9e66f3768c842f37ca0cbfd329eae13f79754cf8e29cbf279e66b24f344be5 +size 9856 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..de33e8de6c62f499995c4105cb664fe5e2c19172 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9f77e3ac6f46daf456492a6789dce6c09e0ee2a7d94cbaef23556e17e70c142 +size 5662 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..442903768ba1167a5b5f692f2f7dcbd1eb8f5d4b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0951175549a5cad668d924737c1b0ef486379852b2a85ac66608f19992ac2097 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..686335c75fadf5d6d14c21bd06b0521e3d10a945 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:adc75d3dc26d02cddbe5e9d37e2ce24d5abbb60ed6bb9e2a97ccf5d11aa97d32 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..e5d814de82dbfc85db27f4a8378200ef70cf92d5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45fd3ad2d728a40a3a1d04eb9fb1658ebfc769ada35c2bcd9cdedb7e3de4957a +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a1eedf4fa19b31a575a903ed1a9912ad988565f3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38f8632252265aefc69e9c4f51673ca68178f9c6dace13e2a630140effbfa0c1 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..51dc63090f9056f9421d0ef583d5320de7c91dbd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7be61bb7c94c365c46858e95f2ba94be947b96c295dadaf347ee3e3d150d8b0c +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3864ff40d45220c5af6b380d61ea9767dd943527 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9fd3d38359837e6dffaf0ddf301d236fff833a0c3a806448d529884eac38f9f +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e27bab4f80bf97cab974954ceb2019ccd5203ec8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ee8727b4e3885a19c95f04ee0cedce1ef28cd0cee38379733a798f9a6bd0f6a +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b8f6f1dc7a65f66897ca8965825c7a8965da9ed7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98f201d2bd8d5764a93037e948b9f97b6393e4a625119fce835a5769457aaac7 +size 6976 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..49c1d5a8960dc07828635ce74d91738df9578a74 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:322244ab15b37379a7b87d196a0f82f633d8055b1052e930c1c3cd90a0272008 +size 4190 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..c1eee7af5b0eac54878c792ea54fd6976711647f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dcf1c414cc60d59e64e401ed38a077d4fb335a30ba07808fc2a306ac124b2ebe +size 50471 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4460a698a015a66871d4450d31d7eda0e2b845d2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f89ffbd6ce2f5ecd4333ccebfa20c13df8e6417d0ddc0479f103f1978238124 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..3d020fa72dec6083dced3f2922409627e5615a6b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d5696f61f52f3fbad1d415ce623d63046e6424c11f684d2b400346cbbe7b21c +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..64865cdfad524bc5470ba6f87a4a3114ce113049 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8d1b9df90b722d557865b7d442246836c92de102dc85809c1af72950c31a643 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..94b24a0d2bc709b73a5982af52a36e213521908b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c764c00cccc4e58fd56029d99f4dc8b45c06ee506114c891724bcf1717940525 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ad62d7462dbea033e51b4856069ca11b62b85dc2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:daea17588761b6d259c5854d4ce2e86ea314676dcb013451ebf754aa00d0ebbe +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c6843331513ad1324cc68852013a39a9d468ba15 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c14e6d6827b7a36028b63235203bb9d0ce17669871c3d863a95c77c77792975 +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0013fe6ccf3b7ae56eec360ba0122aee58083404 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2483dad0c87e0bcb8d8f6d00c7536c82c5fc89e62874328ae91db49436e9ad32 +size 7552 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..281e7ee334d04657234d67cec5c8169c2cdfc2c0 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db9b3699259cbc76c17517b650850237eadfdbfd6dcc44e512d115f89e8da6b4 +size 4510 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..507bc6f0cfea48d2062331a9efd53892d8c60e86 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68481c49474763fe8eff9cd455a910c67a631a0d29312b890bb80356627a5589 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..217bed2265bc52b87a02450bf54e1c99e839d9b3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:13850a7bec2da149ee2df3743d7cc21a7097418e7a7f86a7039cb8b083d644aa +size 50382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..6b735a89530ae133261e00834bf060a0bdf53300 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b696ec04c9d399e74c75cb76471fdde7ac0cbe6ee7ba72792e9df31ab4cc0ad +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..16ec6d01246e7e27f273a0ea8cf352347a8539e9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d9037cf8b915de320d09f98fa2f0e36cff9748fa65645d81e5e435089d33b2e +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..6005f51d917a1fa9fcbd91f0e6076e02d27ee7b5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be66ba59d655eff69af271c2fb7b8c522c8e834a5ada8a7a0b06e688215e8baf +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..29e570556d80c0f5144528f575e6ff6dcbd6aa6b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac5cb5f4c6f8dabf3418c771ad44b00fdeeef1138921d52a439d53a36cb86cc2 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5a71f0206c31c947332c331f288195681d662ced --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4696d1b5c7990f564461b4edf85a42750dbeab9d1cb995ecf5b5dc1b816bbc4 +size 75550084 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..eb7bd8aaa50fedbfea6163492b04b83712d37cc7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe182b1612ee0b4953700a2502e87233c811a5af7449d68f0d57d367a89201fe +size 7360 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..cd81422995abe2ba1c9a85d8f21e6cc400283a03 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2bea718a2281f6540d5c2385f48cb0318443462261ddb086939c5a8b130be8b +size 4382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..7a4cc20cf8b3242844f4d7e6bd4fe940d5527882 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec940b7b52a10d3ddd926e24dc5f7466fe4eff97a1fb89834e78e722d6156fa4 +size 50471 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ba81fc67bf9f34ca3cb5e15d36c722492fd367d1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19d4eedb4c949b208ffc1f171c93af5fd24121d103998092be9ab830dc235510 +size 50382 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..c8fb039bca29a8a71f3cfb8005e7c33aac72d4c9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5945fc40094fb1a6924b5da7014c1e936b83ab8f7ac341d505f26acb5c28a2d6 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a93fbe7b9f4765d7efcefd7201d6897ab569ae55 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9eedac48c71a4e91d0dc1d966825fbc8c78a6b898422493bd4a855ecc6e5e019 +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..09fdced935e106d986d075630bc56e2c3d888dee --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e14d28b986bc4c2fb486a59ce8ae99179318bc3f04c21c1a70ca8d495d104385 +size 50337 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..cb083f3dc3851aeed4d31dc14d77bc37a0d1c5b1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x16/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f583b93966ec80ce3a1faa6d6377457eedc7a86cf6e2f6fa3353c82128033f0a +size 50387 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5fc62717bcbd11c64aa832dc67929181fca3a64d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de966ee7b48f390f12bae8eb31ab8735b1d111471cc34e7441557a1d9a60ed8d +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a55ae345b6c7d4cf7c7d3cad58bd71fb50e5deca --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67b679576e3ec67af6897d061aaaba793e05837c81d72f40806b0983f26991a4 +size 5376 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..892c90c3cfd0942bb8b7e2b5d77b1c137a97bf78 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:262406d01ff7db88c582fcacf1a38e2c09ce14c6b5a373818e68181e7f0429a5 +size 3422 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..e204f31a873f61c42eda26692c9c9ec59ea2d3ee --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a29abd402f18cdf934a2ca2fd83a6fe106d7f1951ae51179f984b7b925af49d +size 13607 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e7cb6811cc78ebadb027649227bd91e2db4a4dc3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d94c10a5d83ecc31f20e2cc088916164af4804a192b4948a93c30691b5ccf13 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..fddf660c2f836acaecc87af690a81e29b1beed4f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0cdc89249bdeb341a4e04bed8ac4fa76fb5d67030c910d3ae3b295e200c6dbe5 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ba2fbeb62696c26198a57ebdf662afa6024faa8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3625efd5a59c8a896453c993a63bec41f875e10ebaa486c655a140453c9f10b8 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..481e0c74f4a41eadd6942f7a62e83cd171e6e0d4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c48cf72dbd5cd8a8791024898523601842a60a0fd2f6b501e1317c985bc58ad +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b41ccae4c5c0c61fc490ea503bbb188d2faa4392 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf05eea566a208cf7f45b5b29628f5417a184a8101b56d48c1819678d3b605a2 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4af27ed9e7b7a6f3a01b1c604bbd2ebb1b044ec2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0513ebcba0bcdeb7dd592cfeba31722ae2a92a81f7f78c33200c38db025982e9 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bed14dc1e70ea48206f0e25c663effb61dcb8f6c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7de65ac51909ef51ab24398fb2f1b2a48cb1b2cf6347df10a2fea7100ba98c76 +size 5376 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..36f9468c34edd93f71a9f3d6d4a4aaf331f5cd71 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a4d38a6beef7221c57560edd6fa6d60e403bd24857cdf9c633707ebed1e4fa3f +size 3358 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..fae3bb4124a98ac4029cadf5e368c106a6018035 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5a98ac83e43f6355ac7199dc38d1f7e749cabf8f510cacc96feeea26a2e0028 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..1dcc098ca3c9a6e117019b3de17666ad48b8c466 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e688a3c53ab39e0dccb8efadbd3b9c67a0632884e38f495688c1422b1cd21168 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..f200c7df0c9b4600100467023c7c2c58f3475a47 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41d3a65b8590d9c3fb5051f961dd9c22ac9384921416dd1fe5e41f7a12ddccdf +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..febefb795d876d911654499a75312b1b346f47e5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6718374a1caf950c6a5fed07a741d17d7180a93759ef5e0a1ee958ef9dbd790 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..fcef88ac25edd09a006752acb928d61fbb6f0203 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b7e82a8ca82df27c74f018032bcb3dc34d4ccd4c940cd5c41453c6f924528b64 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..79e7ff23f28903222939b758a9647c90ac86d751 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d95150c64f739084054c29996234797907a012aed0986269d8777b9db30a0cc +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7cfb520b077676038ef9a040c5451f4f2a987039 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2dc21a20d3236eec79af432d85c4ccc131c229b141036479ad962fdb73e93f80 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..de4ecdb23a72c6370c00dd5a6408694f85f56f32 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fefb62fb01ea00d0677716d72a766eed367dcf7799202f89363955b550b19e7c +size 5440 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3ce7c26097d46b063d01be31025ad147777cfbb6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f305ba5b53ad42e8eb63007b13a502f721716c0756bd22ea51a9f86917147862 +size 3422 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..67a07448c29a1297492b5a5b78faa24168aa2c54 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0cc4a789d4f84a70a27e9358dd399cf49dd5438d112a8bc53cb8145196c6343 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7029b326c647051c159beef4b89b17a18c4d692e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c0e1ef7b9185641c96a5a21462825d4190ae63b1e5842fe31e29466265bda9d +size 13518 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..e4b3f164a376b8da3dbc27b4b893018c39df4b5e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61c45a463fec220a0d7f1c9a0761fbe505ce7536b785872697e6142364047b5e +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..025a2a75b9a3ac3f022f84ab58c26db6ac140edf --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c03e346d7a31fb7a4316b1d068a502ec4d583e29438a5c616c3115abee634446 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..4211a758a5772027ddaa043b5dcd5158674f0a08 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57a3f12eaa596908eb0db4b16deb6ee466a34a46a8a844a906692fcfb89caeac +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..35a6bee7daadfe23a4d3cd9cb7a7dc5f2d93f67e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b1ba902feba52380675bebe5b2738ec8e25b4588197ab590cff348ebec4f30c +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0f78d40ae0f1354072e2e48e0ab8af39defaf3f3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de54780136f1fe884bed953480cd8c4b9f4a591f82a2344cf104d8817142b491 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e44b7a3187ece8151154182f0e8f5852310f9081 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2181cf297d0590cb830f8f0e358bfdfb0bbc5066dece3b653dece65210d24cfb +size 4224 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f079dc4d3599643c6cc04ad5c832c8c412c5f287 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:402ea10e05a100efa8275d25e377f8eac6315b0de6dd35d7f85a72addb8a2acc +size 2782 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b47f671e06efb78c66a604d1d1a1632221f6599 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da243a76e6a3a4da1ad0e6f067fd0c7207b1ff06cbc39545ce72a8f81e815f8e +size 13607 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..25d9e51124e444f4f87b3aca1b60478c8c2811a3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cdbc806de4ca664bad239625dbf49d4dd51aafc7104f3e046e66296dd551be1c +size 13518 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..3cbc4a1199e4c932c6ab6fbdb00d8148eaa8d20d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:199f3dc5e887acb7a3bdfb6e8fcaac53288bde636f71bdbe499dc65b4ad45401 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d72acb31633e572fd03c3f519186b7c1f827cde0 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00f5ff9a05179f6737e12a3db47f9e0bd50418f1d5b789605982d9564515e418 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..6f80551c58a2c2b614ec75e6d1766e785425fd27 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95646c09decc3027fc1c22211da2a49b4266bba1cc0b7ce486f3c49ccb9ba009 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f2254b8693a05f32cebda98d7b1e0193b7c20204 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c53120384949cd1aea3b42f0c8eb968fb0986864668f0eb873b0df410f37a1b6 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ba63e65ce255850fde88a72e4032508fed135d74 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c61015a4f0a0a1bf933ea514a1f2cde7e7c9eed3bf46829ae5c76bf37d7a161 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..afe7b4d781fe226ef679d566a4ee5520823b6fa3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:886a1bc8b4e8e8a8ba0d4717a77da62f040a4384e42e19c007bad3fa1cd74511 +size 4352 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..91fd3dff421c603cc813ad291e2821b2b48a90fb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a18f6e04fb4f73315143aac3f96c8da843a4d13b4954ecbc1363fad94b6ed23 +size 2846 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..af517a00aa61c47e4294ad43705183fb2c21dd6e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28aae9bea92903ca0050cc9c990bb645d6a13251e17bcf7591773a5f806885cb +size 13607 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..47c660564a56eaff7911e56554f9a2f2111e3896 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8eaba49783f748a89c524680600128ba602f67bb58576f6bf5a0f906aef061e1 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..24cf846456c430c25314b43ab54bffa5180d4f11 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec83b45d9aeed1e65d3ae4df429d868fe1125ac079c3240f333a1c003b66cca0 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..6228fc351b13d89f7e8a0de620e5b6c54ba8df08 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fbdb689500d47fa466a5f5d27cfde83b11977b98309080d07d992ee84ae2cc6e +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..b56a5a8fae1b422fa85258c2158a71116f3660f2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b5b45b254f4b5f26b8703460fc287a75fcffa5520a1c04eca4f7afd37ead5fd2 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..972b7d44cf3ca5e30d346bb2abfe42f387c47363 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:999439c6612e93e7e79df8c033dd797a639c14c1cf8732b6b3f880f629f85c2a +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e30bbcb0ef25c680a448ff9fbb9389d0a07af1d5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:55697f468ac8857c4246fc0b02b6d30827d363f1e70d81e852cbd1180ccb95cf +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5a7f38ebb81a0421a6cca62636929471a4eb9f70 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8283e4deb4ebfb21386d43d4b6848d18b396c70bfd06ae71d3ef762d4a6fbb2 +size 4288 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c6a4e2453886fec6c67046f148f14e2e1c160015 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e75623b6ae99a2eba0f8fb71f1bf95d24fd88f73f9702efce7de3659bf4a091a +size 2846 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..fcb8963049c3ad7b5749d042ba85b2c67fa28b10 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e64eff9f3b88224599fe027407eedaeb6b9ee5c0244e67d8d113807398bf50ed +size 13607 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c9113c10808ed8b1cca8140813d94d7e6756f984 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6acf83dc0bbf445f3a118fc47b9ae4642d063de3aee9429c8bceb71453bc10e0 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c07d78a36adc681ce9f60357165f9f2c539daa1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd30605a69b25c24466c25017d2ae610cd1272adf52f87038562c6ff0531b6c3 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..266c1b6d605292db6a04cbb2ac6d5b31e749ac57 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:046c7fe82d4a31a51c7a0c09e14053a5373f117ce6fa511c9b691af60e020d62 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..b04973447a06afd71f7de2fda7f4c66a93970ab3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b571839fba8f9ec983f6227c1b4c0f823ac7f4a98a75e15679868b0debcf331a +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..9fa7cf3258c2029305c777bb8f55cb715850bc39 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7660ed98c15bddbe3929abbb040bf41c8ebf7856e9e568add4c316a55dbe277 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..04d8780210118ea9bd798740a00a18b4b12b5691 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0ab644d58e562e5ec2f915af2a11dbb38ad3a6b6e0921b27aaef579c58a7bed +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0088a361ac7e0f3991827cccf9c73354c1e72624 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:63b7c49e7af7dc6c406ff23da448910dc26ac5c0f18e9d879da5bcc6d3315cc5 +size 5056 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..29c335965e6c1dc4a032cc8192a659504d9efb7c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:550891e976d3e852f15054558fe85377bcad07dbfd212a36056ed86b64110df1 +size 3230 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..f8cb7d87a5621f7c29b0d05e7436bf91ab193e16 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:638b54789706404cb54d677eda9d6fe778e8b8beb5b7111ff5d112902bc3ff77 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a78ddcc6b77dce420c1822c9fb8a2b3628e7ac89 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cb90acb8eb0733c92e2a85ff5a7c095f6c2d47eb7e1711cb199ff88bbcc11a8 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..7c7c366b4c4247e1d01efe6303d906036e87f915 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e06ff7692d1c0247eb6ce0bc392c20e2950377964b3c07b431d0a471f26020e9 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..31ecd0b7536e2bd7082fa3b5d4778ad4bc6e0209 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5fcff04da25636e497c3208ecfd8f06533a1eda2ecc9be37e7a6adc3e6fda4ad +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..a8480e6decb0d7863e9f82fc6ad540c28650323f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1bd76d60b2831e7d4dcc88f4ce5faaf3a50b9f217c9d52238f56e4987d668656 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4a5b1a4544a90386f0e7b28efe803e5b12f9b11c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a90728688444c555c1e171150d0951198b10e4de9885f1352c4231d9a786e0c +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..04d10d8143ac465552ea97d2728bb112eb38548d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82c0990d53a03de28d4faacdc1dd028116304d242fd1bcb901ad4d118907d9cc +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ac370534a2204d40227d30ebb05dd76e10b4d1a7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9b16d126a5a9b8a5f7d2bada90a5ac73599fa674b6fee045298b62e5aece758 +size 5376 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4d28ac71612f5432ac31d115cacd6cec7f5a8a9f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4dddd16aafa96bbb84a8bea783e399cf97fbc86fa7f72ccb7a6080d4e251696f +size 3358 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..6b52c5f6187f9c5f8ac98e0941271f05e067f8af --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c476193d4ca7f27976c9d8282dd812d2f4c6820816a22ee715fea16e5d8c899 +size 13607 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..ccd3cb598e0f035b33ff7c4843f4f75d3d7e9fa6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02bacc26c6b4e166a48126a11c14bf822015dc112bf07472e2f090bbb5c2013d +size 13518 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..458db414c6201890f6c1c2a7427d70e0639cb58f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d1426a84a75d9796f6d36e17ae0558ef3c767a8247fe412ddc70515c3dde531 +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5d8951fa8d7977e7e19e36f1c56be91256f4c0ac --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3458c9c1e2a3652c48d151029e985fdc72c3c3cdb2d1abbc648e183df8f3a5ff +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..061a61fac338f333aa033a338d05d3930880b149 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c80c189fbeccdb1729ea913c7ed5f303376ad08ce4faf8a3c05ebc115d85151e +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4c50698ca12fce3e3be24f07c7bed69b00ac0979 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7a18b420bf23fce5c4461e2706633efbe0e7e99fb805c8ebfdf57db7608a310 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7786c164ed4e4f0f3fb1deb0f0c2dd8ae02b286d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3aa375fac916c5b841d6da40e975066903be0746508e37378023ce4acd52a158 +size 18890116 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d949027f1d695874de056d5b2719a43274056fdb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53ebd5b1d4eda5104638c6dfef3f68e99d031406ce440746efafcc2f15b7ce81 +size 5120 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..1150a26ab4e6bc1e9e0e5139874ec282627dd8ac --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61fac5893a4f525726e07dd244762324d486f15414ed8e93bb9a022c5d37aeb7 +size 3294 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..576f32cdc950d2abf7a6dad7a83ea83b030bf6df --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79674d13614c40ac3d47d846b7a5773e32481beb8062502dbf7fc5c247501f1d +size 13607 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..587ef4924f1b3eb8ede8114597698a65a1883432 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bece1a91b2ffb95d4a1de9036740d654974431ecdeec7c30e476c44339bdb4d0 +size 13518 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..014196bafdffb32442d8d4b7ef246223887d76e5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b1f5616aae4ada9b3c423212d86e12d5c160202915b4b5ab7acdd4462ef38cf +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3abd40354ce693d778aed62f5458790dca4b6b0e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0faebe72c3b41ba093c0a4ec93e8fb2ae1371aaa2107447cf770279af07bf1f6 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..60f5f913321c026ab9e61e565b61031068c97028 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9876dcc8ebb900442ab7d91a02a3ce20bd1fab3785f6076e0d04bd7023662ae +size 13473 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..3dd75a0977ba82b0af31f483c158c9c9d8c14f76 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x4/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6c2c374792605a000a0c48f8a3712991073022b8d17e413fbde0e6fa9eebbe5 +size 13523 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..eac423349c022d255906876224f812af77b022bc --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8517e5b7369a0c70236c01866659b6364055c56201da3e0fe9b9cbd909eaf582 +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..323ad3c6cf40e334172ae3546ba002f8357b3f06 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:705bec0646b0afa4148c062bed5abe0bbd739ab72e8eaff769442b93e5d7934f +size 6208 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2b41e0183de0cec3ce1a9aaf406ac248ee5aedf3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cfc4baa15f8e9ba34f5aee9ed8f4aeefd4f177c3212a8ce940263749419619a0 +size 3806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..77a760a357ff528d429a6d6448256b8b5af3ba9d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a4e59f8536c1f4a6e59f42ad18892d63ed33718b21406a3f5fc6d4d67f5d109 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..0726a0fc2272d0cf26ba2154a3b498a741cc25d9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7a64ce5a5dfb397b88d6dd9ba2bf1d9a3839e26208febc527c4a54f2435e040 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..d564ef27f5bc067225b2b8edbf3136da9ec16791 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e168c9ef01a1cbabe16d4f506a03af9ea964d8af73522105721316af44f2bf9 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..53a8cc2d0b63c689e2393f2a434ce1838db457e5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f86f9e679aea3c804f2365ed3f7a53aa32bea86499df5ccc99f39b73a2bae52 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..8687edd4638a70e8c8e9deae90f85d7a525b843f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57471fdd671d85d50ca05b80daabd349434bac1b54079fb575cced7c4f2f1a1c +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2607b8a31747948ab2bcf699b49f9a93cf6da54a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e165a56943e05e68b95d6c3021feb72c032b2c5e3510a42dac88a4e6c1ad1552 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3dcb9563befb71e2a0e4b834732919a3ed356320 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67868a452c701fe18566d59f55e05fb89d5c5f4e5fd7621b56048bd1782b9dde +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b80551fee70abab0077f6c31b0882a6e8a9d304e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:826c386b2b8e6be4e203958c44c40e987ab0ed8a22750547af0db969491ef58d +size 6272 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e68dc4d29fd57ce9eaacf09ac1b7de18458b54be --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c04e5ea50e2091425386fab0dcb4c826651908acd4379fa981170ad4c1e2242 +size 3870 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..6856e9bb1c129af2fd671b124a944342dc9dec32 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05e9b5da4379753eb400d6f3743c008d8e9e2e4deb98bebc88d28b11ff0cb9d5 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..68a81d399b5e47a9a6c8451908dfbafacc94d867 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6b88ca7fdffec20c8932f67ed7fa8de172e0df1f0cd6e0dde9b7e3fa8247cc2 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..71de9410dddd9ff34f663fe457299e8ca1e3b94b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8fd00799ca904b92aae306b78147995c3e92693e4ea8535a6344990a156369b3 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b3c7fd9cc29d40aec7c8e6eb6b55cab682ad247 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83622c54e4d4f0df0f0eb28daea77c049843e74a7e0f0eb475149fefd906f606 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..1c77b20882c3096ca39a4c387ef0ec450cf01b66 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:085726299b910c2cddd4359e06eda7f1ef1af27d429645a490fb12548fed6477 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..549c4fcf6acdd6b944408e58648f35c7bc9cf654 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2838587faf2ad91374575fa7c8a6a5d5b9a6516308ab0aa9829d7b55e24bd76b +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3c715c6d5dd88b29fd9b3d97587b21336c630efe --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80725e38cae1808d8a6423556090deefabc2afad4f3951aa124e52dadd5451ce +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5995cb9c34713e9d68f6d4737961226788308e84 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a1e48ba5a980de3135f2948a3a12e05a1959ad2c11ddbf152133ea2732ec92e +size 6336 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..300b72e9162fea3b361b6a131ed7e39c9acab73a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:007be09fb2ce68dbe2f2407d7920c8db8a671926c311e74f2c3415908ecb2714 +size 3870 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..a36a123c9ab497fcd44673fce021b09696aadeb5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5923c7421061e4f506db12787e17c517ae5e6f087eed731688c8a7103199cfa6 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..1df3ee1fc37cdf67d39ca0e69ba4a24325070248 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fac3e4ed242bc20a33330ecace724bddbb98d50d43ac4369057399c1998e6dee +size 25806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..6faa429e2c2593c674f3c8dd81784217e28f73fa --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e1d8c78cf8b8a589bf7b352cf152e6be79c2744647f24c2c275c50256bf3fb99 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..c26213f7369fa7d0c8be478f18102bf7956c9a5b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:287df10a30d5aaf072a49605a38ddc55a7153dea668b15918f0dcc9e83705af0 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..3cc9c7efca073a20a14c85452e6409524dcd77f1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aadede3a20d07ce6e6606dbb7fcdce94dc950cd5bfd3e50c2386f4fdb7f5e4a5 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..81bbd8833fe09451b0f9fa22d8b2b36ad51a349d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:feab31a263229e4f24b1b5952aa4a09af1013300bf23ac59e8762f1fc7fdf6fd +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f7b23195b4914ee1c207db68e67466e4d642527d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc097d2bfa5ff8a95cad46e883780b58629dbfc88af2f3b96d0730864ada5da4 +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d72c5aa1013095bd3b73fd55602382ea8fd0697d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ec39e33394ab9278c8289def48fb22dc7c2f7e5441d0087b52488dce570abd2 +size 4928 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..04f1afcebe8e038a455029a00d60419872594110 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:878e737f030b46f5ac9d842e176c73fd6805dbfcf4c77f94e9d9915495295a2e +size 3166 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..a245c1410f2449f548491a03d928485d5f97e947 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e155f8268a0b0eb0583cbf20221eaef1f6deabe9bcad08d6d9edd4a66efc620c +size 25895 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..dd5b595e22dcea3d3e5083dff014cb55ca770b31 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:749128de55b33a06139a9c4ad586d161061d344ca5ecce1875faf5d75897897b +size 25806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..52b866950cb640a4da9da532c0b60c031eb423bc --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf5de548c92024ffed2b6300aa19a37cefe4e05d35031c0a8ea480c429f97f87 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..50c8ffa1b8e76660c3e4c9d40c9158e163b1f8ed --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7852951d22e5ede1fdbf2283a84a49baeca3e3ea692a1747e13cc2ec536daef1 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..97ea7daf905091fdc484b5b33bd710de1e8b1186 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4da03171253ee8b7937915d07097cf75739e9a6e89ee05fb554766c84721c07c +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..372a71ebecc325ef75aaf700d5f7bf408b91a14e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a8664a7876015f9ecdff5c6fc4f5f41db3d3bd0cd4ca2c9cdbfc96a8bc3827f +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..64ce7481569be4e56411229583a560b54d58e6c2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84ef1d7936b1e13c3ba1285fea725b136e3a27106438d564504d2e00c4df0aa4 +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..6fa0dc5f27c44e22bfd06b57989997ca50268857 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ce273e6a50d2204620f9eb25f63f55d4181bce68af959aa29b8be2ab11a2e494 +size 4928 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5d08f716ef3067805a224c8d1a616bc2005bfd6c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82785fd409c15f16ba9e071519f49ce2b0878c36915b1a1e33167eb82f581059 +size 3166 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..fc81178f5290a27997c945b98aea511e95fc549e --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eda4eb2fc219c6420997d49dee46fb2e0c364368887c64c46759fe11ae939928 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..03cbab825d28287fba59d64cb89aaf5b58d47953 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65ec6136306f23108771245af04787c0d389f6ddc5ab944c0fcc4a42a19b2c90 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..e8fa074da314b70468453b9e759bf153e49fb081 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ecb50eaeafbdd76d81fa5f48819e664a33f03a606ac9f63af41383b404cfff29 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8af07bc709225915326e4bdfa9e13b7414b0b5ba --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:544468395a25ed455018f5ac0cb877f0984aa14e7f366e67efb05ecbc63b9c64 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..788d76d847ec95ac6f1384e125a7a3878389af01 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a6475a62d7301f3e23f9a974b00f4218c8caede7e2aa4d8f39f72f352211507 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..eb193522f6f6383c12de555022c08be50253afbd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08df41d3871bb41f1bcfd5c86d8f05c563866561f5b07b045aba7ff255c54986 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0aa91e3127308580cf87748eba151f4d2b42998c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95826fea462ac093e520f6bc8adc1e3703860dde0b982cea0a9a131d9f9c1e54 +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f526931c237f6166e4a98445874d2cc684e5a0ca --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8d2a93e1a7e29d6e86245d3cf237cb152cb75d33da840f34da96c35a4b4de65 +size 5120 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..011e0866e94ea7240bbb283bc965c8313ed96a6c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fc0c972e7ec15e3efd91b92b3b64994898aa9e8892fdc11cb63c86b35d6563dd +size 3294 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..068f1d70b2b0068b36e902290ef64c1e0b8feafc --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a58dd822934b1565f088d0949364f0967ff892b2bff9254af0a1c35c7a4884c +size 25895 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..76db66407ae891754096abf1a836e5929c259d7c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb656db60f16b9fb419be7797b844e64c9e2e2a2389c726b19216d7f6195331a +size 25806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..4249210eeff378e2e81496e4b9daaf809fa1a94c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d45d0fd4b58808ea23d7f8f607d900aea93d428e96ee05e4532cb7ccd7562665 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2e2c1108b3be54c329e29f4b541332bc81dd1469 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5761b0ea6d3e23512fc63b1f3eabb70072e8deadfc094d7a1ecdc07af9cb06da +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..fb8e3b10b902a8a0012ef1590619b04b53b4affe --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:870d3212736bb949d8112f6eff928533a58b8d8161c3a2e3f7ac87596091986a +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b054dd8e93d2c0cb44049b57a154d8345f7f8aa1 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:158d4046c584352904cc459aa2a2469d65954307a528dccb9d496c6d15001c24 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..19dd6d893fa42b5646ece55a2cd2e925c0133695 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f60e46a6679c69300aede556d7562ba70859ff743c93c63343ce904600fde70e +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..16afecd113454eaa3135b43fb59d7af326cd2b58 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:440a7859f3be52ee5c63b2d96b817e21b6cf13e6ba5e57f9480d6da6d3ddf697 +size 6400 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e8e599a004338653f3bfa69ca624dbb70963440d --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:771d0cee0c7f6925a6bd5ac1853188a994eb895ae51001e358fa98f7a1101f5c +size 3870 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..a9c96a1f35e11ea090619a9a4509fa865a0402f5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0d73aceae77773dfc655461f48013935b1bd37a9686f47272fefeea805a8c9c +size 25895 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..eb7d20831c0e57f301b643a294c457de01feb57c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:63d09230657ab0504aabf1ef48c23e63a207851412f559f92fab893461508cb8 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..0f556a7549dacd8c8c61f0de5644488feb8170b9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ce4d4aecfed022f0d0cf58bffd04ab246c49f156000d77864706927cb48fa56a +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8cce6dc2cb124360e7b5544d87c9c5bd23955fa0 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d628720f866cae6eae809a0b7098ca7193bf578fe315c2e0fd18375e401d91b5 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..1f93bcc7e9211643004a3451a82116973745e3e8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71a7cb059c4fcc62132cd5b3da81a1712e47eef23459020e2b66438537cb6fcd +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..92fe772f8b26f17b01a8afecf5a610702f999c9b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d4f43d40a89061fc61f66bde2f93d00a061d1c557902c682e1bea26a4f7e7e +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bb2e63606d3b6782aa87d8da1525f196e1ed32b7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e449c0be7ea80b98b919ac3abec5a1ab44dd54619f3687169c77040ddbebae9d +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..81d219e063c831a7c21429b15eb55db5187b1596 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2eaf92f73126fe3cd7321354e1a6b05d2836675125a6ac92784a32223acc2970 +size 6080 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..752447f14b357828a1ba3cd416de8722b1bb5d71 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2407d2882f6c69855534fc2799d7de01cab55c43bceaa38db66d7bd663b22fd +size 3742 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..0b8e79ecf38361b18abc1dad2201e1dab74c1d64 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9763948e2d95b5e59472de85f9ec84bcd3674acffdd7cdb9520efb119ad5ee87 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..01f403d26fec9530f37f673d7ecb9e5d9a8d1790 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:922122d363bcd825ab7d6e66b30482dc9d1497a33dcaffef0c293c6e57f3a34d +size 25806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..7de616d9fe996d754970fe3e1753271bcfdd3779 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d0fc5da76e18cb7606450955a96dbd22f926f7fcab9d603800337e574c3405ff +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..a2dec354b03967cddf11b6384e411b16e26311b3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:07b8acc845a08b858fb269e84b0d0200370af996f2f08637290b039844d16c1c +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..df534b374c9ec9f7742fcd0fda5fc17425ebd0cd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4008c80fe438fbffdf2d6575d2d8583b4ded5bf1dd76fa6255bef9e843ac7c38 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b2891566e7bf8eba055718e525f0b629b7949835 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71a07a979977b1891df5f880613bca8a7da87225e26ada63fa5a046663b5ea68 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3e6250d750b2739035e39d94e58dc74d103a9ddd --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3edfb31815f6cf1b6d0825e3ed4550b4159502eff24096bc4569bc16b87721d4 +size 37776772 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..d65374f8d7824120080f3dd0cd68bb1fc3215a41 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9fbd72da61989a61e23854811d3c30e53ef1f995d4312177238e56d9ecdf0dca +size 6272 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..07fe74786cf4eaf8f4a43708c5417786ef7b8f14 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a4cf1c1e3aa174a374fe01f045857677c3d174c14522a92c02caaf1dce7b312 +size 3806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..e3594c3eed8341fb2ead286df2a70bbf319f83f5 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a4fda166d739edbf88c03ae48a26a282ee15f8bbcb8391de27c36c45a71796bd +size 25895 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..91391aa76452041b33470bf8c40de119540bcd8c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f28dd37e2a20b567f05a39df186d5b88234e409ab673b18cac35465fc5cfbe1 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..bc91cbaca241e0d0780b1be02c12d78f0e7c3e95 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:07e8fa15fa86a90ce9a26c94a03b6700abdba8eb7b678b8c3dd9002f57af5ef4 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..7f443a656aaa8ec28c0014a9b1b9b1ee8753600a --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cceeec9a2419f227844758c0f64b69a6e8693020386118056e4ae4aff8124c17 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..a014b83373e4fd175e1e67c7836ebfe4dcf5f54f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f9525f9cff082d5affe508266a09d4b0a1a1a79e4ce23b1a90eda8b2b5845d73 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..4325bee01f86bc397c76849f211d7c63c34adeb2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/matryoshka/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14e305076317809cd58bb534c0e08fa2acae194529722d029228805ca92afcec +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0addb4dc9a8a34593e8532d53a77ecd9b1fc15c7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7851c061771a023ea962f79c81f912b4d030df4767f1782dd67cd5127fe801f4 +size 37776696 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b5925e24d51859bfcaa8d991068eff5d55e48a77 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9369ebe396873b8076197e9a24a9b89b06cbda9506547d8668845c3daa1c7135 +size 1664 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..80d8a2c9d119247845fd429a0649a9007dcf25a9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be77a12dd61335302340885944b4bafce2fd0b9ed2e430962817c8452c4f3fa2 +size 1502 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..6119aeba5e10cc05251924ccbe9884f8e45b65b3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b2836437090f493ae0c6e7af8c5e5fe35047ea1bd404db6604e4984901a2f55 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f83c7b1b142ce0b45e2b90a3534659cab3043848 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d07d9f7b9ec088a6e3243c7ae7b957f01156067968911529f0ee8bdb15487fd3 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..adbb69edbb9a806831a773ccd9774c2c3a84a42b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25c1a00ceb148b838c621d694a122020fa9f8d5d8beeadc074530a8af37c1ee4 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..fb7b11986b579271c3ef7c47d5fb24a8f37a81e6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ec687d6a8f28385aa849bdc59e7ee8ed559a61de8e3fbc41e9e4ee26ff40524 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..b058347929d3454a6d44fbfaf7a76edb11af7a68 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7b7d776bd74ea13ccc1b0e514c19d60f28580b8e8b95726d19b8565cac709d5 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f13a66fc805e2386fcc271117b37430e21d4ad3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c275b50db7b5ba0ace2110791981bf4028e4da5135ad160395e6e37f23a618d +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..818e8467411f248069f84b170fb9cb42344bd632 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab4cda46458d4c566ef14c060334b0cf429a06e52fb73f8e1716fca8ca5d9d6e +size 37776696 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2a5d86a81934c02cdca18958dd76c9d3d3222c47 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7a817bd0dca0486a1e373c60a0bbaa3497cc82cd1ba19da9226bde12950d889 +size 1664 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c6e96ed77cebe5616c9cfc23a6d30728d3d47a4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf99a003b34b5f01f19a5e2935a5d28b8e6e1a752c227b2a99f9becb8e2267c5 +size 1502 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..3cdae94711ae36badf74bc2f9abbe7e3cc1ecc4c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4aea9f3d62c60458f7d053489ca2c2bafde65d1fbe6b27065badf5ea5728ca0a +size 25895 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f52e2eabaf0b32c35724bae23bca3ce0508b1ca0 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9feda350f1787a3cdf9f779f38b882830e3e4639f732809ee323c6946d258673 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..5202c66b0909e82a723d232ea488317b5feb1b1c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7549b7fef9a271324c39a1c25b3db0635b1b0eb5d1e28ef21421a87ae334beda +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e3f7b893412d609485983ba81fc235fe86c87471 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0fb4e89445dc4bb3f0b2cceb1eb6921550d0745cdd7e52ee8b1ee32500c98267 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..04bc4d3e5625a25400effc42e15ffa85f9384282 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef0c44487972b8c10d7373b920dec3d71d42f5fb1e1c9c8bbc9ba97d9a1e8a7c +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..34b0cfe3813f54d930fa61e4280f0e9bf5895dc2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68f8b75406080e267057714101f0817639ad1068b94de12e01be5e63a5ad18bb +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0874ebf3ce77fdceb5a06902d786828ad95ede8f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c1d9024ecdf19b5e98948e0955452835f08690d7d7651d26fb949ae9f3cf1761 +size 37776696 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..48f8f91d1752202c0cf7b58ea94a245daa44290b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b80e0cdae9db614a8e11a15facfb3601f172cad5a34cfb7dfae0de8376c48c3 +size 1600 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..fc0868be33ebd0e21d9f068566c0903d936862f3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d639427ba656b944e26c28227696d1372d9b8173ac166c8eaeec9d5717159759 +size 1502 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..a5d1d93b655e910d2743ed405d5b3c783402c141 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c5156dc9e4529bad167e53d5ec871dc065eca574a71d8e9ceac7b53c3e32d2a +size 25895 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bb21a959a29683fac64dfc53bcd70863e8ec85c8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8b9e44b3b9ad6a5bff9b2c45c4bce921c43d45f8907b0f1e233651385cb0aa05 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..ada58e5dda6cb7e4ae53919bccf5ba4db0848cdb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7937f5b94b8753c3bef42fea8af50f4277bb3ee4f3f071e5ee5b95eba76d133 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8e2bbaeacb2375a03ade0e7556959ccf83f6b0f4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c45b5135bdd898d7b460803f596486b0691c13caac0eeb4dc87cb2b7ca74ce4 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..4f61b7b61473fd646efadc3f9a655bf7acfa9b02 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6e21ef28a626b5d933cba801ea362692c439a99ec78babc98871281098bad9e +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..238b5ab364fb1d03df3646e9079c7deaa906fa38 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_128/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ede9121d10c2ecd3326a3615d8e1805f2d2331f47deab4747a128c5a108f9f59 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/model.safetensors b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0addb4dc9a8a34593e8532d53a77ecd9b1fc15c7 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7851c061771a023ea962f79c81f912b4d030df4767f1782dd67cd5127fe801f4 +size 37776696 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b5925e24d51859bfcaa8d991068eff5d55e48a77 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9369ebe396873b8076197e9a24a9b89b06cbda9506547d8668845c3daa1c7135 +size 1664 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..80d8a2c9d119247845fd429a0649a9007dcf25a9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be77a12dd61335302340885944b4bafce2fd0b9ed2e430962817c8452c4f3fa2 +size 1502 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..6119aeba5e10cc05251924ccbe9884f8e45b65b3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b2836437090f493ae0c6e7af8c5e5fe35047ea1bd404db6604e4984901a2f55 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f83c7b1b142ce0b45e2b90a3534659cab3043848 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d07d9f7b9ec088a6e3243c7ae7b957f01156067968911529f0ee8bdb15487fd3 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..adbb69edbb9a806831a773ccd9774c2c3a84a42b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25c1a00ceb148b838c621d694a122020fa9f8d5d8beeadc074530a8af37c1ee4 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..fb7b11986b579271c3ef7c47d5fb24a8f37a81e6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ec687d6a8f28385aa849bdc59e7ee8ed559a61de8e3fbc41e9e4ee26ff40524 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..b058347929d3454a6d44fbfaf7a76edb11af7a68 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7b7d776bd74ea13ccc1b0e514c19d60f28580b8e8b95726d19b8565cac709d5 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f13a66fc805e2386fcc271117b37430e21d4ad3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c275b50db7b5ba0ace2110791981bf4028e4da5135ad160395e6e37f23a618d +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/model.safetensors b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..818e8467411f248069f84b170fb9cb42344bd632 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab4cda46458d4c566ef14c060334b0cf429a06e52fb73f8e1716fca8ca5d9d6e +size 37776696 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2a5d86a81934c02cdca18958dd76c9d3d3222c47 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7a817bd0dca0486a1e373c60a0bbaa3497cc82cd1ba19da9226bde12950d889 +size 1664 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c6e96ed77cebe5616c9cfc23a6d30728d3d47a4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf99a003b34b5f01f19a5e2935a5d28b8e6e1a752c227b2a99f9becb8e2267c5 +size 1502 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..3cdae94711ae36badf74bc2f9abbe7e3cc1ecc4c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4aea9f3d62c60458f7d053489ca2c2bafde65d1fbe6b27065badf5ea5728ca0a +size 25895 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f52e2eabaf0b32c35724bae23bca3ce0508b1ca0 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9feda350f1787a3cdf9f779f38b882830e3e4639f732809ee323c6946d258673 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..5202c66b0909e82a723d232ea488317b5feb1b1c --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7549b7fef9a271324c39a1c25b3db0635b1b0eb5d1e28ef21421a87ae334beda +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..e3f7b893412d609485983ba81fc235fe86c87471 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0fb4e89445dc4bb3f0b2cceb1eb6921550d0745cdd7e52ee8b1ee32500c98267 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..04bc4d3e5625a25400effc42e15ffa85f9384282 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef0c44487972b8c10d7373b920dec3d71d42f5fb1e1c9c8bbc9ba97d9a1e8a7c +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..34b0cfe3813f54d930fa61e4280f0e9bf5895dc2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68f8b75406080e267057714101f0817639ad1068b94de12e01be5e63a5ad18bb +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/model.safetensors b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0874ebf3ce77fdceb5a06902d786828ad95ede8f --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c1d9024ecdf19b5e98948e0955452835f08690d7d7651d26fb949ae9f3cf1761 +size 37776696 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..48f8f91d1752202c0cf7b58ea94a245daa44290b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b80e0cdae9db614a8e11a15facfb3601f172cad5a34cfb7dfae0de8376c48c3 +size 1600 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..fc0868be33ebd0e21d9f068566c0903d936862f3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d639427ba656b944e26c28227696d1372d9b8173ac166c8eaeec9d5717159759 +size 1502 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..a5d1d93b655e910d2743ed405d5b3c783402c141 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c5156dc9e4529bad167e53d5ec871dc065eca574a71d8e9ceac7b53c3e32d2a +size 25895 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bb21a959a29683fac64dfc53bcd70863e8ec85c8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8b9e44b3b9ad6a5bff9b2c45c4bce921c43d45f8907b0f1e233651385cb0aa05 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..ada58e5dda6cb7e4ae53919bccf5ba4db0848cdb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7937f5b94b8753c3bef42fea8af50f4277bb3ee4f3f071e5ee5b95eba76d133 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8e2bbaeacb2375a03ade0e7556959ccf83f6b0f4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c45b5135bdd898d7b460803f596486b0691c13caac0eeb4dc87cb2b7ca74ce4 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..4f61b7b61473fd646efadc3f9a655bf7acfa9b02 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6e21ef28a626b5d933cba801ea362692c439a99ec78babc98871281098bad9e +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..238b5ab364fb1d03df3646e9079c7deaa906fa38 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_32/seed2/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ede9121d10c2ecd3326a3615d8e1805f2d2331f47deab4747a128c5a108f9f59 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..b5925e24d51859bfcaa8d991068eff5d55e48a77 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9369ebe396873b8076197e9a24a9b89b06cbda9506547d8668845c3daa1c7135 +size 1664 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..80d8a2c9d119247845fd429a0649a9007dcf25a9 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be77a12dd61335302340885944b4bafce2fd0b9ed2e430962817c8452c4f3fa2 +size 1502 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt new file mode 100644 index 0000000000000000000000000000000000000000..6119aeba5e10cc05251924ccbe9884f8e45b65b3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/firing_rate_per_latent.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b2836437090f493ae0c6e7af8c5e5fe35047ea1bd404db6604e4984901a2f55 +size 25895 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..f83c7b1b142ce0b45e2b90a3534659cab3043848 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d07d9f7b9ec088a6e3243c7ae7b957f01156067968911529f0ee8bdb15487fd3 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..adbb69edbb9a806831a773ccd9774c2c3a84a42b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25c1a00ceb148b838c621d694a122020fa9f8d5d8beeadc074530a8af37c1ee4 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..fb7b11986b579271c3ef7c47d5fb24a8f37a81e6 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ec687d6a8f28385aa849bdc59e7ee8ed559a61de8e3fbc41e9e4ee26ff40524 +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt new file mode 100644 index 0000000000000000000000000000000000000000..b058347929d3454a6d44fbfaf7a76edb11af7a68 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7b7d776bd74ea13ccc1b0e514c19d60f28580b8e8b95726d19b8565cac709d5 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f13a66fc805e2386fcc271117b37430e21d4ad3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed0/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c275b50db7b5ba0ace2110791981bf4028e4da5135ad160395e6e37f23a618d +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..34b0cfe3813f54d930fa61e4280f0e9bf5895dc2 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed1/stats.imagenet.split_val.seed_no/sparsity_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68f8b75406080e267057714101f0817639ad1068b94de12e01be5e63a5ad18bb +size 25811 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..48f8f91d1752202c0cf7b58ea94a245daa44290b --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_hist_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b80e0cdae9db614a8e11a15facfb3601f172cad5a34cfb7dfae0de8376c48c3 +size 1600 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..fc0868be33ebd0e21d9f068566c0903d936862f3 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/active_count_proportion_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d639427ba656b944e26c28227696d1372d9b8173ac166c8eaeec9d5717159759 +size 1502 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..bb21a959a29683fac64dfc53bcd70863e8ec85c8 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/max_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8b9e44b3b9ad6a5bff9b2c45c4bce921c43d45f8907b0f1e233651385cb0aa05 +size 25806 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt new file mode 100644 index 0000000000000000000000000000000000000000..ada58e5dda6cb7e4ae53919bccf5ba4db0848cdb --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7937f5b94b8753c3bef42fea8af50f4277bb3ee4f3f071e5ee5b95eba76d133 +size 25761 diff --git a/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt new file mode 100644 index 0000000000000000000000000000000000000000..8e2bbaeacb2375a03ade0e7556959ccf83f6b0f4 --- /dev/null +++ b/vision/clip-vit-b-16_layer10/mp/x8/l0_64/seed2/stats.imagenet.split_val.seed_no/mean_act_per_token.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c45b5135bdd898d7b460803f596486b0691c13caac0eeb4dc87cb2b7ca74ce4 +size 25811