Qwen90M step 880000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 875000,
+  "step": 880000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "9f61ed48f2cb4c3f7a06a34a94d9060f253db6eb2c73e5fe5e4c3640c5f78d50",
+    "model.safetensors": "4b9d2a70139a0d1f9df3b8d0e12559fd9c5a968e5f392e046f198e787d4d4f9e",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "a291745e8cb415b9662fa7db6054a73b2972d734926e2a0eeb0e4f105d90a9b2",
+    "trainer.pt": "c9cc764700c82332c48b3399ebf24c59ed1c14aebdba6e49722271643ac53c92",
     "training_config.json": "0ab64376a29947d9a9f670c0ece06c1900ab6f9ec82482957af5fc2c70beec58",
     "performance.json": "172134497605e50805257548c05018674e515323c63930c6237cd46d43786367",
-    "shard_scheduler.json": "4d5b87679412b97e1700b3b9d9840dca98fd012ebe6d7c8389acbcbf734e9db5",
-    "coverage.json": "54ea23c51a1da54ebc5dce287846cffb5ef7a00a7d2bf8f7eed2277d17896c85",
+    "shard_scheduler.json": "ab4e6115c9bbc89a09b88644bbc2cb0f60e1b63c69cfcc06c7e28677dd166d5c",
+    "coverage.json": "b3a51956dfe1df698095be4a7fc4872ba4b95279d28c7c7a86b1b4833aa682ce",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 875000,
+  "step": 880000,
   "stats": {
-    "climbmix/blocks": 31500000,
-    "climbmix/targets": 52657150038,
-    "climbmix/input_tokens": 52660486950,
-    "ultra_qa/blocks": 11812500,
-    "ultra_qa/targets": 19271353866,
-    "ultra_qa/input_tokens": 19271390893,
-    "cortex/blocks": 11812500,
-    "cortex/targets": 449143762,
-    "cortex/input_tokens": 18776747701,
-    "ultra_style/blocks": 11812500,
-    "ultra_style/targets": 18684178798,
-    "ultra_style/input_tokens": 18685158198,
-    "openmath/blocks": 7875000,
-    "openmath/targets": 12168335207,
-    "openmath/input_tokens": 12172389842,
-    "rewrite6/blocks": 3937500,
-    "rewrite6/targets": 6455009145,
-    "rewrite6/input_tokens": 6462085998
+    "climbmix/blocks": 31680000,
+    "climbmix/targets": 52958296496,
+    "climbmix/input_tokens": 52961652321,
+    "ultra_qa/blocks": 11880000,
+    "ultra_qa/targets": 19381474490,
+    "ultra_qa/input_tokens": 19381511739,
+    "cortex/blocks": 11880000,
+    "cortex/targets": 451709207,
+    "cortex/input_tokens": 18884030914,
+    "ultra_style/blocks": 11880000,
+    "ultra_style/targets": 18790932141,
+    "ultra_style/input_tokens": 18791917045,
+    "openmath/blocks": 7920000,
+    "openmath/targets": 12237932422,
+    "openmath/input_tokens": 12242010271,
+    "rewrite6/blocks": 3960000,
+    "rewrite6/targets": 6491849239,
+    "rewrite6/input_tokens": 6498966787
   },
   "sources": {
     "climbmix": {
-      "page": 4314,
-      "row": 579,
+      "page": 4338,
+      "row": 2798,
       "start": {
-        "file": 208,
+        "file": 209,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 68,
+          "row_group": 81,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 17670723
+      "documents": 17771246
     },
     "rewrite6": {
-      "page": 4,
-      "row": 876,
+      "page": 18,
+      "row": 1453,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 16384
+          "row": 73728
         }
       },
       "epoch": 49,
-      "offset": 2048,
-      "documents": 10082056
+      "offset": 0,
+      "documents": 10139977
     },
     "openmath": {
-      "page": 2696,
-      "row": 1147,
+      "page": 2711,
+      "row": 2854,
       "start": {
         "file": 25,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 126,
-          "row": 568
+          "row_group": 188,
+          "row": 8
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 11043963
+      "documents": 11107110
     },
     "ultra_qa": {
-      "page": 1165,
-      "row": 3053,
+      "page": 1172,
+      "row": 1635,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 30655
+          "row": 59327
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 4774893
+      "offset": 2048,
+      "documents": 4802147
     },
     "ultra_style": {
-      "page": 1779,
-      "row": 128,
+      "page": 1789,
+      "row": 636,
       "start": {
         "file": 10,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 12636
+          "row": 53596
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 7286912
+      "offset": 0,
+      "documents": 7328380
     },
     "cortex": {
-      "page": 229085,
-      "row": 121,
+      "page": 230394,
+      "row": 107,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 29323001
+      "documents": 29490539
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 875000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "9f61ed48f2cb4c3f7a06a34a94d9060f253db6eb2c73e5fe5e4c3640c5f78d50", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "a291745e8cb415b9662fa7db6054a73b2972d734926e2a0eeb0e4f105d90a9b2", "training_config.json": "0ab64376a29947d9a9f670c0ece06c1900ab6f9ec82482957af5fc2c70beec58", "performance.json": "172134497605e50805257548c05018674e515323c63930c6237cd46d43786367", "shard_scheduler.json": "4d5b87679412b97e1700b3b9d9840dca98fd012ebe6d7c8389acbcbf734e9db5", "coverage.json": "54ea23c51a1da54ebc5dce287846cffb5ef7a00a7d2bf8f7eed2277d17896c85", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000875000"}
+{"step": 880000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "4b9d2a70139a0d1f9df3b8d0e12559fd9c5a968e5f392e046f198e787d4d4f9e", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "c9cc764700c82332c48b3399ebf24c59ed1c14aebdba6e49722271643ac53c92", "training_config.json": "0ab64376a29947d9a9f670c0ece06c1900ab6f9ec82482957af5fc2c70beec58", "performance.json": "172134497605e50805257548c05018674e515323c63930c6237cd46d43786367", "shard_scheduler.json": "ab4e6115c9bbc89a09b88644bbc2cb0f60e1b63c69cfcc06c7e28677dd166d5c", "coverage.json": "b3a51956dfe1df698095be4a7fc4872ba4b95279d28c7c7a86b1b4833aa682ce", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000880000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9f61ed48f2cb4c3f7a06a34a94d9060f253db6eb2c73e5fe5e4c3640c5f78d50
+oid sha256:4b9d2a70139a0d1f9df3b8d0e12559fd9c5a968e5f392e046f198e787d4d4f9e
 size 360622744

shard_scheduler.json

@@ -22,8 +22,8 @@
       "tokens_per_second": 598315.063954628,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 75597.86980031131,
-      "required_pages_per_second_upper_estimate": 0.005230036044010423
+      "required_token_positions_per_second": 76089.26603843171,
+      "required_pages_per_second_upper_estimate": 0.005264031975959954
     },
     {
       "source": "rewrite6",
@@ -36,8 +36,8 @@
       "tokens_per_second": 8533.957213384612,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 9449.733725038914,
-      "required_pages_per_second_upper_estimate": 0.003956406177794145
+      "required_token_positions_per_second": 9511.158254803964,
+      "required_pages_per_second_upper_estimate": 0.00398212334571673
     },
     {
       "source": "openmath",
@@ -50,8 +50,8 @@
       "tokens_per_second": 380144.35761751566,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 18899.467450077827,
-      "required_pages_per_second_upper_estimate": 0.0039362606573777655
+      "required_token_positions_per_second": 19022.31650960793,
+      "required_pages_per_second_upper_estimate": 0.003961846876730323
     },
     {
       "source": "ultra_qa",
@@ -64,8 +64,8 @@
       "tokens_per_second": 43742.16556775435,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28349.201175116745,
-      "required_pages_per_second_upper_estimate": 0.001699491323724437
+      "required_token_positions_per_second": 28533.474764411894,
+      "required_pages_per_second_upper_estimate": 0.0017105382440332022
     },
     {
       "source": "ultra_style",
@@ -78,8 +78,8 @@
       "tokens_per_second": 22018.86276523721,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28349.201175116745,
-      "required_pages_per_second_upper_estimate": 0.0022040256964075268
+      "required_token_positions_per_second": 28533.474764411894,
+      "required_pages_per_second_upper_estimate": 0.0022183521574413653
     },
     {
       "source": "cortex",
@@ -92,8 +92,8 @@
       "tokens_per_second": 910315.184299418,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 22,
-      "required_token_positions_per_second": 28349.201175116745,
-      "required_pages_per_second_upper_estimate": 0.3636331136737182
+      "required_token_positions_per_second": 28533.474764411894,
+      "required_pages_per_second_upper_estimate": 0.36599677741963155
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.12715467205484288,
-    "producer_p95_seconds": 0.31598223599939956,
-    "gpu_update_seconds": 0.9752655755347274,
-    "producer_batches_per_second": 7.864437726430466,
-    "gpu_batches_per_second": 1.0253617323175905,
-    "producer_headroom": 7.669915385524231,
+    "producer_batch_seconds": 0.4706437804607049,
+    "producer_p95_seconds": 0.3458738019980956,
+    "gpu_update_seconds": 0.9689671597405202,
+    "producer_batches_per_second": 2.1247492084589275,
+    "gpu_batches_per_second": 1.032026720356333,
+    "producer_headroom": 2.0588122056813654,
     "sustainable": true,
-    "buffer_seconds": 60.466465683153096,
+    "buffer_seconds": 60.07596390391225,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 86,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 484630.29120586516,
-    "upload_bytes_per_second": 4285666.65230903,
+    "generation_bytes_per_second": 228212.41980255014,
+    "upload_bytes_per_second": 3851625.6327998103,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 103.99037887699095,
+    "estimated_full_shards_per_hour": 48.96907277638796,
     "selection_policy": "Smallest shard within 90% of measured local peak, enlarged to amortize 30 seconds of measured page production; bounded to 256 MiB."
   },
   "upload_measurements": [
     {
-      "seconds": 1.6203178860014305,
-      "bytes": 6721657,
-      "pages": 29,
-      "shards": 3
-    },
-    {
-      "seconds": 1.8514023890020326,
-      "bytes": 12113021,
-      "pages": 32,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0979482909970102,
-      "bytes": 1151793,
+      "seconds": 1.605746043002,
+      "bytes": 4717702,
       "pages": 35,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.1090724049972778,
-      "bytes": 6289086,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.3987950150003599,
+      "bytes": 11634070,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.7820321440012776,
-      "bytes": 7365760,
-      "pages": 31,
+      "seconds": 1.6460893040020892,
+      "bytes": 7447219,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.5119859270016605,
-      "bytes": 1025135,
-      "pages": 32,
+      "seconds": 0.9642833869984315,
+      "bytes": 493177,
+      "pages": 16,
       "shards": 1
     },
     {
-      "seconds": 1.464499204997992,
-      "bytes": 17226254,
-      "pages": 41,
-      "shards": 5
+      "seconds": 1.69654296200315,
+      "bytes": 11528790,
+      "pages": 47,
+      "shards": 3
     },
     {
-      "seconds": 0.9209507179984939,
-      "bytes": 949705,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.0500477669993415,
+      "bytes": 2444431,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.9317821920012648,
-      "bytes": 6082541,
-      "pages": 37,
+      "seconds": 1.1570196089996898,
+      "bytes": 6331951,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.9298742110004241,
-      "bytes": 11871053,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.2861647640020237,
+      "bytes": 6463355,
+      "pages": 21,
+      "shards": 2
     },
     {
-      "seconds": 1.752713240999583,
-      "bytes": 1247797,
-      "pages": 36,
-      "shards": 1
+      "seconds": 1.3757250710004882,
+      "bytes": 11835703,
+      "pages": 44,
+      "shards": 4
     },
     {
-      "seconds": 1.4635059559986985,
-      "bytes": 7719999,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.5062627519982925,
+      "bytes": 6228251,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.2099752760004776,
-      "bytes": 6423731,
-      "pages": 22,
+      "seconds": 1.0129257990010956,
+      "bytes": 2288801,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.4774333919995115,
-      "bytes": 10519718,
+      "seconds": 1.3661311769974418,
+      "bytes": 11305021,
       "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.5275958259990148,
-      "bytes": 7252352,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.2991957800004457,
+      "bytes": 5765469,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.0232120440014114,
-      "bytes": 1138613,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.2353659159998642,
+      "bytes": 8486468,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.0960563930020726,
-      "bytes": 6111629,
-      "pages": 38,
+      "seconds": 1.681274881997524,
+      "bytes": 6344982,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.3281459399986488,
-      "bytes": 11209095,
-      "pages": 33,
-      "shards": 4
-    },
-    {
-      "seconds": 1.3028533390024677,
-      "bytes": 6867913,
+      "seconds": 1.540335444999073,
+      "bytes": 2545551,
       "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.275760786000319,
-      "bytes": 7467840,
-      "pages": 35,
+      "seconds": 1.1612642559994129,
+      "bytes": 9600274,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.235418092001055,
-      "bytes": 6152144,
-      "pages": 36,
+      "seconds": 1.406910452999,
+      "bytes": 5890041,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3146963539984426,
-      "bytes": 5075870,
-      "pages": 30,
+      "seconds": 1.0961665050017473,
+      "bytes": 2428850,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.278188054999191,
-      "bytes": 7607915,
-      "pages": 40,
+      "seconds": 6.228057981999882,
+      "bytes": 11900976,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.058052360000147,
-      "bytes": 885421,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.5676271939992148,
+      "bytes": 12360812,
+      "pages": 44,
+      "shards": 4
     },
     {
-      "seconds": 1.6197210989994346,
-      "bytes": 11971576,
-      "pages": 38,
+      "seconds": 1.7888847590002115,
+      "bytes": 1223080,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 4.01536119299999,
+      "bytes": 5846921,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 2.7826384430009057,
+      "bytes": 7726579,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 3.1373434490014915,
-      "bytes": 11875917,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.4537843559992325,
+      "bytes": 4298944,
+      "pages": 20,
+      "shards": 2
     },
     {
-      "seconds": 1.013686755002709,
-      "bytes": 986186,
-      "pages": 27,
+      "seconds": 1.740884680999443,
+      "bytes": 961930,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.333218751999084,
-      "bytes": 7645187,
-      "pages": 39,
-      "shards": 3
+      "seconds": 2.1066386819984473,
+      "bytes": 13346856,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.3886132959996758,
-      "bytes": 6063422,
-      "pages": 34,
+      "seconds": 1.6676214010003605,
+      "bytes": 6129871,
+      "pages": 39,
       "shards": 2
     },
     {
-      "seconds": 1.2067982320004376,
-      "bytes": 5015589,
-      "pages": 35,
+      "seconds": 1.2406462889994145,
+      "bytes": 5768030,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.4427611729988712,
-      "bytes": 13297372,
-      "pages": 36,
-      "shards": 4
+      "seconds": 2.4877858859981643,
+      "bytes": 7529407,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.798427575002279,
-      "bytes": 1107567,
-      "pages": 33,
+      "seconds": 1.8479857339989394,
+      "bytes": 6111417,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.1358039750011812,
-      "bytes": 5944518,
-      "pages": 25,
+      "seconds": 1.0848596139985602,
+      "bytes": 2384316,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.3701994980001473,
-      "bytes": 11970289,
-      "pages": 41,
+      "seconds": 2.864481454998895,
+      "bytes": 16026880,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 1.0856028650014196,
-      "bytes": 1171092,
-      "pages": 35,
+      "seconds": 1.1737483289980446,
+      "bytes": 1071557,
+      "pages": 32,
       "shards": 1
     },
     {
-      "seconds": 3.5694688390030933,
-      "bytes": 7359997,
-      "pages": 31,
+      "seconds": 1.4824117819953244,
+      "bytes": 7847757,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.4348496300008264,
-      "bytes": 6710244,
-      "pages": 28,
+      "seconds": 1.698374239000259,
+      "bytes": 5912160,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.629695798998,
-      "bytes": 10655327,
-      "pages": 38,
+      "seconds": 1.2937959129994852,
+      "bytes": 7777018,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.3922439659982047,
-      "bytes": 7460601,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.521280608001689,
+      "bytes": 6127630,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.1743250350009475,
-      "bytes": 1256562,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.3758144730018103,
+      "bytes": 6889142,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.874451825999131,
-      "bytes": 7588108,
-      "pages": 32,
+      "seconds": 1.881114518000686,
+      "bytes": 7983801,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.605746043002,
-      "bytes": 4717702,
-      "pages": 35,
+      "seconds": 1.3452542180020828,
+      "bytes": 6178324,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.3987950150003599,
-      "bytes": 11634070,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.2824458319955738,
+      "bytes": 4898067,
+      "pages": 41,
+      "shards": 2
     },
     {
-      "seconds": 1.6460893040020892,
-      "bytes": 7447219,
-      "pages": 36,
+      "seconds": 1.4413858800035086,
+      "bytes": 7671053,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.9642833869984315,
-      "bytes": 493177,
-      "pages": 16,
+      "seconds": 1.4289778950042091,
+      "bytes": 1006990,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.69654296200315,
-      "bytes": 11528790,
-      "pages": 47,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0500477669993415,
-      "bytes": 2444431,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.5516541759934626,
+      "bytes": 17080870,
+      "pages": 28,
+      "shards": 5
     },
     {
-      "seconds": 1.1570196089996898,
-      "bytes": 6331951,
+      "seconds": 1.6339488890007487,
+      "bytes": 6146671,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2861647640020237,
-      "bytes": 6463355,
-      "pages": 21,
-      "shards": 2
+      "seconds": 1.0191766480056685,
+      "bytes": 906488,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.3757250710004882,
-      "bytes": 11835703,
-      "pages": 44,
-      "shards": 4
+      "seconds": 1.433525214997644,
+      "bytes": 7753087,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.5062627519982925,
-      "bytes": 6228251,
-      "pages": 37,
+      "seconds": 2.9534675750037422,
+      "bytes": 5286995,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0129257990010956,
-      "bytes": 2288801,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.3232240569996065,
+      "bytes": 8048155,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.3661311769974418,
-      "bytes": 11305021,
-      "pages": 40,
+      "seconds": 3.3732278300012695,
+      "bytes": 11426245,
+      "pages": 26,
       "shards": 3
     },
     {
-      "seconds": 1.2991957800004457,
-      "bytes": 5765469,
-      "pages": 28,
-      "shards": 2
+      "seconds": 2.9584781199955614,
+      "bytes": 1202687,
+      "pages": 36,
+      "shards": 1
     },
     {
-      "seconds": 1.2353659159998642,
-      "bytes": 8486468,
-      "pages": 39,
-      "shards": 3
+      "seconds": 2.338399190994096,
+      "bytes": 11297335,
+      "pages": 36,
+      "shards": 5
     },
     {
-      "seconds": 1.681274881997524,
-      "bytes": 6344982,
-      "pages": 32,
+      "seconds": 2.0653895860014018,
+      "bytes": 6188816,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.540335444999073,
-      "bytes": 2545551,
+      "seconds": 1.478956710001512,
+      "bytes": 1024661,
+      "pages": 32,
+      "shards": 1
+    },
+    {
+      "seconds": 1.288585857997532,
+      "bytes": 7415692,
       "pages": 34,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.1612642559994129,
-      "bytes": 9600274,
-      "pages": 33,
+      "seconds": 1.3701110699985293,
+      "bytes": 11305656,
+      "pages": 28,
       "shards": 3
     },
     {
-      "seconds": 1.406910452999,
-      "bytes": 5890041,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.8134901779994834,
+      "bytes": 7862983,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.0961665050017473,
-      "bytes": 2428850,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.568131148997054,
+      "bytes": 886900,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 6.228057981999882,
-      "bytes": 11900976,
-      "pages": 33,
-      "shards": 3
+      "seconds": 2.2302413529978367,
+      "bytes": 6385571,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.5676271939992148,
-      "bytes": 12360812,
-      "pages": 44,
+      "seconds": 2.218881031003548,
+      "bytes": 11370155,
+      "pages": 33,
       "shards": 4
     },
     {
-      "seconds": 1.7888847590002115,
-      "bytes": 1223080,
+      "seconds": 1.2318224780028686,
+      "bytes": 1293663,
       "pages": 36,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 4.01536119299999,
-      "bytes": 5846921,
-      "pages": 29,
+      "seconds": 1.8874604160009767,
+      "bytes": 5227490,
+      "pages": 13,
       "shards": 2
+    },
+    {
+      "seconds": 1.7371134120039642,
+      "bytes": 13601006,
+      "pages": 50,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a291745e8cb415b9662fa7db6054a73b2972d734926e2a0eeb0e4f105d90a9b2
-size 721771711
+oid sha256:c9cc764700c82332c48b3399ebf24c59ed1c14aebdba6e49722271643ac53c92
+size 721787007