Qwen90M step 785000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 780000,
+  "step": 785000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "861f90278f68313b09eb4b302450aa4b5f8e1fe53d075de34cb4897b13546b54",
+    "model.safetensors": "9e9efa160bc5d9cdcfdc2ca5084b0841b10448f0013b892e4c614e0c0ff5502f",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "bb7ceaeda93bfb3b23d8a7a9fc6b379ef73c40abcbb8d58fb638329c569f90e7",
+    "trainer.pt": "da99d6bb0dd16c05dc0dcdcb4b5f86f1c1ca2eda0fbea8eec1ea8b1c34c0fc81",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d",
-    "shard_scheduler.json": "060f57d01a3356f0f8395d272cc24eb06656aeba401c5dcbc708fe202e44878a",
-    "coverage.json": "f641f693c6260b4b353bf8683e17a82b7b292ad28deab6832a52269599d71b92",
+    "shard_scheduler.json": "70243b4426905888cedd0ae2bc3df6ea021387e3c4d1a15ff1b998d804d4625b",
+    "coverage.json": "2153e069f9b927c191b3d524d2beb884e93a0e0cf00c2acbc8c4eeeff4ac0602",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 780000,
+  "step": 785000,
   "stats": {
-    "climbmix/blocks": 28080000,
-    "climbmix/targets": 46940530276,
-    "climbmix/input_tokens": 46943505263,
-    "ultra_qa/blocks": 10530000,
-    "ultra_qa/targets": 17178444425,
-    "ultra_qa/input_tokens": 17178477489,
-    "cortex/blocks": 10530000,
-    "cortex/targets": 400402885,
-    "cortex/input_tokens": 16738077716,
-    "ultra_style/blocks": 10530000,
-    "ultra_style/targets": 16655737210,
-    "ultra_style/input_tokens": 16656610869,
-    "openmath/blocks": 7020000,
-    "openmath/targets": 10847110567,
-    "openmath/input_tokens": 10850724850,
-    "rewrite6/blocks": 3510000,
-    "rewrite6/targets": 5754125788,
-    "rewrite6/input_tokens": 5760434426
+    "climbmix/blocks": 28260000,
+    "climbmix/targets": 47241568234,
+    "climbmix/input_tokens": 47244562296,
+    "ultra_qa/blocks": 10597500,
+    "ultra_qa/targets": 17288525662,
+    "ultra_qa/input_tokens": 17288558958,
+    "cortex/blocks": 10597500,
+    "cortex/targets": 402969786,
+    "cortex/input_tokens": 16845487108,
+    "ultra_style/blocks": 10597500,
+    "ultra_style/targets": 16762084445,
+    "ultra_style/input_tokens": 16762964029,
+    "openmath/blocks": 7065000,
+    "openmath/targets": 10916644933,
+    "openmath/input_tokens": 10920282460,
+    "rewrite6/blocks": 3532500,
+    "rewrite6/targets": 5791040560,
+    "rewrite6/input_tokens": 5797389432
   },
   "sources": {
     "climbmix": {
-      "page": 3845,
-      "row": 3683,
+      "page": 3870,
+      "row": 2266,
       "start": {
-        "file": 186,
+        "file": 187,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 10,
+          "row_group": 27,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 15752803
+      "offset": 2048,
+      "documents": 15853786
     },
     "rewrite6": {
-      "page": 37,
-      "row": 3740,
+      "page": 1,
+      "row": 3122,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 151552
+          "row": 4096
         }
       },
-      "epoch": 43,
-      "offset": 2048,
-      "documents": 8987664
+      "epoch": 44,
+      "offset": 0,
+      "documents": 9044994
     },
     "openmath": {
-      "page": 2403,
-      "row": 2100,
+      "page": 2418,
+      "row": 3723,
       "start": {
         "file": 22,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 236,
-          "row": 388
+          "row_group": 297,
+          "row": 828
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 9844788
+      "documents": 9907851
     },
     "ultra_qa": {
-      "page": 1039,
-      "row": 799,
+      "page": 1045,
+      "row": 3516,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 34212
+          "row": 58788
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 4256543
+      "offset": 4096,
+      "documents": 4283836
     },
     "ultra_style": {
-      "page": 1586,
-      "row": 561,
+      "page": 1596,
+      "row": 2099,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 77044
+          "row_group": 4,
+          "row": 32666
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 6496817
+      "documents": 6539315
     },
     "cortex": {
-      "page": 204216,
-      "row": 101,
+      "page": 205528,
+      "row": 6,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 26139749
+      "documents": 26307590
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 780000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "861f90278f68313b09eb4b302450aa4b5f8e1fe53d075de34cb4897b13546b54", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "bb7ceaeda93bfb3b23d8a7a9fc6b379ef73c40abcbb8d58fb638329c569f90e7", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d", "shard_scheduler.json": "060f57d01a3356f0f8395d272cc24eb06656aeba401c5dcbc708fe202e44878a", "coverage.json": "f641f693c6260b4b353bf8683e17a82b7b292ad28deab6832a52269599d71b92", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000780000"}
+{"step": 785000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "9e9efa160bc5d9cdcfdc2ca5084b0841b10448f0013b892e4c614e0c0ff5502f", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "da99d6bb0dd16c05dc0dcdcb4b5f86f1c1ca2eda0fbea8eec1ea8b1c34c0fc81", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d", "shard_scheduler.json": "70243b4426905888cedd0ae2bc3df6ea021387e3c4d1a15ff1b998d804d4625b", "coverage.json": "2153e069f9b927c191b3d524d2beb884e93a0e0cf00c2acbc8c4eeeff4ac0602", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000785000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:861f90278f68313b09eb4b302450aa4b5f8e1fe53d075de34cb4897b13546b54
+oid sha256:9e9efa160bc5d9cdcfdc2ca5084b0841b10448f0013b892e4c614e0c0ff5502f
 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": 77198.30405065721,
-      "required_pages_per_second_upper_estimate": 0.005340757798968437
+      "required_token_positions_per_second": 77070.43158003519,
+      "required_pages_per_second_upper_estimate": 0.0053319112847457955
     },
     {
       "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": 9649.788006332152,
-      "required_pages_per_second_upper_estimate": 0.0040401647277631784
+      "required_token_positions_per_second": 9633.803947504399,
+      "required_pages_per_second_upper_estimate": 0.004033472536117103
     },
     {
       "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": 19299.576012664304,
-      "required_pages_per_second_upper_estimate": 0.004019592719392171
+      "required_token_positions_per_second": 19267.607895008798,
+      "required_pages_per_second_upper_estimate": 0.004012934603540484
     },
     {
       "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": 28949.364018996454,
-      "required_pages_per_second_upper_estimate": 0.001735470169819424
+      "required_token_positions_per_second": 28901.4118425132,
+      "required_pages_per_second_upper_estimate": 0.001732595510057987
     },
     {
       "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": 28949.364018996454,
-      "required_pages_per_second_upper_estimate": 0.0022506857176818108
+      "required_token_positions_per_second": 28901.4118425132,
+      "required_pages_per_second_upper_estimate": 0.0022469576468795716
     },
     {
       "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": 23,
-      "required_token_positions_per_second": 28949.364018996454,
-      "required_pages_per_second_upper_estimate": 0.37133135823035174
+      "required_token_positions_per_second": 28901.4118425132,
+      "required_pages_per_second_upper_estimate": 0.37071627919746025
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.39113858623451847,
-    "producer_p95_seconds": 0.258143648999976,
-    "gpu_update_seconds": 0.9550468874500143,
-    "producer_batches_per_second": 2.5566385807827743,
-    "gpu_batches_per_second": 1.047069011103749,
-    "producer_headroom": 2.4417097189112105,
+    "producer_batch_seconds": 0.09094359758591963,
+    "producer_p95_seconds": 0.24977909400331555,
+    "gpu_update_seconds": 0.9566314666790962,
+    "producer_batches_per_second": 10.995826276338393,
+    "gpu_batches_per_second": 1.0453346297205295,
+    "producer_headroom": 10.518953418082141,
     "sustainable": true,
-    "buffer_seconds": 60.1679539093509,
+    "buffer_seconds": 60.267782400783055,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 82,
+    "upload_pages_per_commit": 84,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 310654.2999911742,
-    "upload_bytes_per_second": 4935605.042243135,
+    "generation_bytes_per_second": 337241.253587939,
+    "upload_bytes_per_second": 4932306.221605968,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 66.65918111611767,
+    "estimated_full_shards_per_hour": 72.36412244537952,
     "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": 2.3685029429998394,
-      "bytes": 11410659,
-      "pages": 34,
+      "seconds": 1.8041386129989405,
+      "bytes": 11898323,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 0.8315555410008528,
-      "bytes": 1017212,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.2189035779992992,
+      "bytes": 6954573,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.2100102580006933,
-      "bytes": 5596380,
-      "pages": 21,
+      "seconds": 1.5599068500014255,
+      "bytes": 2415324,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2646026780003012,
-      "bytes": 13750693,
-      "pages": 48,
-      "shards": 4
+      "seconds": 0.996803005997208,
+      "bytes": 6353091,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.3428173240008618,
-      "bytes": 5558514,
-      "pages": 31,
+      "seconds": 1.149405411000771,
+      "bytes": 4840926,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.273476110000047,
-      "bytes": 7479055,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.5049040789999708,
+      "bytes": 13497203,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 1.084104961999401,
-      "bytes": 947822,
-      "pages": 26,
-      "shards": 1
+      "seconds": 2.2471259019985155,
+      "bytes": 6049971,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1121349749992078,
-      "bytes": 6275461,
+      "seconds": 1.1292916710008285,
+      "bytes": 2686677,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2975029250010266,
-      "bytes": 6264701,
-      "pages": 26,
+      "seconds": 1.3082805390004069,
+      "bytes": 10490550,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.4136529820007127,
-      "bytes": 11858220,
+      "seconds": 1.171707494999282,
+      "bytes": 6154958,
       "pages": 37,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.0470604990005086,
-      "bytes": 6246532,
-      "pages": 37,
+      "seconds": 1.3786574580008164,
+      "bytes": 2401890,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0089250809996884,
-      "bytes": 2580280,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.469864372000302,
+      "bytes": 11763636,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.1785594149987446,
-      "bytes": 10228291,
-      "pages": 34,
+      "seconds": 1.3772630089988525,
+      "bytes": 6008251,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 0.9172461540001677,
-      "bytes": 2354642,
-      "pages": 28,
+      "seconds": 1.3044562169998244,
+      "bytes": 6256652,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1485085579988663,
-      "bytes": 12167168,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.3504138789976423,
+      "bytes": 6609036,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.0151017729986052,
-      "bytes": 6037243,
-      "pages": 35,
+      "seconds": 0.9880936410008871,
+      "bytes": 2162409,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.2390309599995817,
-      "bytes": 6886865,
-      "pages": 35,
+      "seconds": 1.2899976480002806,
+      "bytes": 11585751,
+      "pages": 45,
       "shards": 3
     },
     {
-      "seconds": 1.3525706070013257,
-      "bytes": 5675921,
+      "seconds": 1.4168642419972457,
+      "bytes": 6615057,
       "pages": 28,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.2551854249995813,
-      "bytes": 7559192,
+      "seconds": 1.1043323689991666,
+      "bytes": 7549149,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.825367109000581,
-      "bytes": 1141341,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.1122719240011065,
+      "bytes": 6287535,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.5828043500005151,
-      "bytes": 15814359,
+      "seconds": 1.1806445659967721,
+      "bytes": 6303271,
       "pages": 33,
-      "shards": 4
-    },
-    {
-      "seconds": 1.3585308149995399,
-      "bytes": 7486305,
-      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.8041142309994029,
-      "bytes": 1148050,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0195367809992604,
-      "bytes": 6121053,
-      "pages": 30,
-      "shards": 2
-    },
-    {
-      "seconds": 2.6576393979994464,
-      "bytes": 1839915,
-      "pages": 12,
+      "seconds": 1.2656237250012055,
+      "bytes": 5785840,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.339241851999759,
-      "bytes": 10779600,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.1340223679981136,
+      "bytes": 1097457,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 4.318171066001014,
-      "bytes": 1094935,
-      "pages": 34,
-      "shards": 1
+      "seconds": 3.1717748339979153,
+      "bytes": 13032899,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 1.3115034270012984,
-      "bytes": 13301880,
+      "seconds": 1.520787964000192,
+      "bytes": 10348218,
       "pages": 34,
-      "shards": 5
+      "shards": 3
     },
     {
-      "seconds": 1.1869826189995365,
-      "bytes": 6445004,
+      "seconds": 1.0010391289979452,
+      "bytes": 2627126,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1019184790002328,
-      "bytes": 6210962,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.991184726000938,
-      "bytes": 5824172,
+      "seconds": 1.0633560690002923,
+      "bytes": 6091672,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.041803035999692,
-      "bytes": 1241755,
-      "pages": 35,
+      "seconds": 0.8510588639983325,
+      "bytes": 1121057,
+      "pages": 36,
       "shards": 1
     },
     {
-      "seconds": 1.3904252129996166,
-      "bytes": 12223668,
-      "pages": 40,
+      "seconds": 1.2576819179994345,
+      "bytes": 11872557,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 1.4052468339996267,
-      "bytes": 11659745,
+      "seconds": 1.5458234039979288,
+      "bytes": 11492771,
       "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.587534140999196,
-      "bytes": 2763901,
-      "pages": 37,
+      "seconds": 1.4007872010006395,
+      "bytes": 2588923,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.0879035409998323,
-      "bytes": 6026132,
-      "pages": 28,
+      "seconds": 1.1760680010011129,
+      "bytes": 5785195,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.9889250640007958,
-      "bytes": 952891,
-      "pages": 29,
-      "shards": 1
+      "seconds": 1.2628126730014628,
+      "bytes": 9956766,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.1961971269993228,
-      "bytes": 12036916,
-      "pages": 46,
-      "shards": 4
+      "seconds": 1.1540077350000502,
+      "bytes": 2614009,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1770325620018411,
-      "bytes": 6029556,
-      "pages": 30,
+      "seconds": 1.9888146960001905,
+      "bytes": 6170248,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.1039014870002575,
-      "bytes": 6886174,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.596614274996682,
+      "bytes": 11247081,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.8041386129989405,
-      "bytes": 11898323,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.2773266180010978,
+      "bytes": 7212495,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.2189035779992992,
-      "bytes": 6954573,
-      "pages": 37,
+      "seconds": 1.216828146003536,
+      "bytes": 6096262,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.5599068500014255,
-      "bytes": 2415324,
-      "pages": 28,
+      "seconds": 1.38459085499926,
+      "bytes": 2407911,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.996803005997208,
-      "bytes": 6353091,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.9793150009973033,
+      "bytes": 10539605,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.149405411000771,
-      "bytes": 4840926,
-      "pages": 32,
-      "shards": 2
+      "seconds": 0.917002105001302,
+      "bytes": 895297,
+      "pages": 29,
+      "shards": 1
     },
     {
-      "seconds": 1.5049040789999708,
-      "bytes": 13497203,
-      "pages": 34,
+      "seconds": 1.181815096999344,
+      "bytes": 13310041,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 2.2471259019985155,
-      "bytes": 6049971,
-      "pages": 37,
+      "seconds": 1.608177831996727,
+      "bytes": 5937731,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1292916710008285,
-      "bytes": 2686677,
-      "pages": 37,
+      "seconds": 1.4116140810001525,
+      "bytes": 4972441,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.3082805390004069,
-      "bytes": 10490550,
-      "pages": 32,
+      "seconds": 2.1865444229988498,
+      "bytes": 7544891,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.171707494999282,
-      "bytes": 6154958,
+      "seconds": 1.1021929470007308,
+      "bytes": 6736997,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3786574580008164,
-      "bytes": 2401890,
+      "seconds": 1.0818150289996993,
+      "bytes": 2336300,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.469864372000302,
-      "bytes": 11763636,
+      "seconds": 1.2755690529993444,
+      "bytes": 16950690,
+      "pages": 39,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0866290939993632,
+      "bytes": 1216935,
       "pages": 36,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.3772630089988525,
-      "bytes": 6008251,
+      "seconds": 1.176057447999483,
+      "bytes": 7472528,
       "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.3044562169998244,
-      "bytes": 6256652,
-      "pages": 37,
+      "seconds": 1.2658942909984034,
+      "bytes": 6109615,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.3504138789976423,
-      "bytes": 6609036,
+      "seconds": 1.894670263001899,
+      "bytes": 6431784,
+      "pages": 35,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2564982580006472,
+      "bytes": 6256940,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9880936410008871,
-      "bytes": 2162409,
-      "pages": 24,
+      "seconds": 1.1927087240001129,
+      "bytes": 6721845,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.2899976480002806,
-      "bytes": 11585751,
-      "pages": 45,
+      "seconds": 1.142808556996897,
+      "bytes": 7320475,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.4168642419972457,
-      "bytes": 6615057,
-      "pages": 28,
+      "seconds": 1.333093295001163,
+      "bytes": 10300175,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.1043323689991666,
-      "bytes": 7549149,
+      "seconds": 1.2003908199985744,
+      "bytes": 2637875,
       "pages": 38,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.1122719240011065,
-      "bytes": 6287535,
-      "pages": 37,
+      "seconds": 2.519623430998763,
+      "bytes": 6025157,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1806445659967721,
-      "bytes": 6303271,
-      "pages": 33,
+      "seconds": 1.0435017019990482,
+      "bytes": 1045686,
+      "pages": 30,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3271499310030777,
+      "bytes": 10074712,
+      "pages": 25,
       "shards": 3
     },
     {
-      "seconds": 1.2656237250012055,
-      "bytes": 5785840,
-      "pages": 28,
-      "shards": 2
+      "seconds": 2.2123866580004687,
+      "bytes": 13632293,
+      "pages": 49,
+      "shards": 4
     },
     {
-      "seconds": 1.1340223679981136,
-      "bytes": 1097457,
-      "pages": 35,
+      "seconds": 1.2271638120000716,
+      "bytes": 936260,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 3.1717748339979153,
-      "bytes": 13032899,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.1434297949999745,
+      "bytes": 7632762,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 1.268450359999406,
+      "bytes": 10380594,
+      "pages": 38,
+      "shards": 3
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bb7ceaeda93bfb3b23d8a7a9fc6b379ef73c40abcbb8d58fb638329c569f90e7
-size 721774783
+oid sha256:da99d6bb0dd16c05dc0dcdcb4b5f86f1c1ca2eda0fbea8eec1ea8b1c34c0fc81
+size 721781183