Qwen90M step 815000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 810000,
+  "step": 815000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "b860c45edec3c7186db36858be040ec7573c74f1e3787dd00557ac89a4199fc4",
+    "model.safetensors": "fdf8df44c98676caf03f6204fff629f95101410723ef61ce23662a551a8f427b",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "a24063ea8ec1a680b2cbad4348b11f3720f7b12a48fa11c042f0aa687a24bfbb",
+    "trainer.pt": "125e3addd8b07f3024e8ed1e3c3e2f705e99c3cfcfa6b9f1547db31a2081c525",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7",
-    "shard_scheduler.json": "d3fc1112c30fbcd245d56533cfa5eb69c877b6ac83ebcc2954147694a2f34a08",
-    "coverage.json": "284e9a6216e7c4741fa1f7e01c706d28eb52bfba351b9120d3853f5e4505d59b",
+    "shard_scheduler.json": "d8c00298782e3f8a28914751297b8d31ce17e96db9eaa5de1147c633462ccb8c",
+    "coverage.json": "1e90d10c39719e43af2b2ea2d77129d1f5aa16ce1b06969bdac4ca8d9ac3d0a6",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 810000,
+  "step": 815000,
   "stats": {
-    "climbmix/blocks": 29160000,
-    "climbmix/targets": 48745844785,
-    "climbmix/input_tokens": 48748933709,
-    "ultra_qa/blocks": 10935000,
-    "ultra_qa/targets": 17839350326,
-    "ultra_qa/input_tokens": 17839384717,
-    "cortex/blocks": 10935000,
-    "cortex/targets": 415808421,
-    "cortex/input_tokens": 17381926974,
-    "ultra_style/blocks": 10935000,
-    "ultra_style/targets": 17296108093,
-    "ultra_style/input_tokens": 17297015211,
-    "openmath/blocks": 7290000,
-    "openmath/targets": 11264198535,
-    "openmath/input_tokens": 11267951300,
-    "rewrite6/blocks": 3645000,
-    "rewrite6/targets": 5975471100,
-    "rewrite6/input_tokens": 5982022644
+    "climbmix/blocks": 29340000,
+    "climbmix/targets": 49047052329,
+    "climbmix/input_tokens": 49050160293,
+    "ultra_qa/blocks": 11002500,
+    "ultra_qa/targets": 17949369763,
+    "ultra_qa/input_tokens": 17949404365,
+    "cortex/blocks": 11002500,
+    "cortex/targets": 418371734,
+    "cortex/input_tokens": 17489228968,
+    "ultra_style/blocks": 11002500,
+    "ultra_style/targets": 17402557074,
+    "ultra_style/input_tokens": 17403469802,
+    "openmath/blocks": 7335000,
+    "openmath/targets": 11333868401,
+    "openmath/input_tokens": 11337644586,
+    "rewrite6/blocks": 3667500,
+    "rewrite6/targets": 6012349385,
+    "rewrite6/input_tokens": 6018941143
   },
   "sources": {
     "climbmix": {
-      "page": 3993,
-      "row": 2480,
+      "page": 4018,
+      "row": 620,
       "start": {
-        "file": 193,
+        "file": 194,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 23,
+          "row_group": 41,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 16357808
+      "offset": 8192,
+      "documents": 16458348
     },
     "rewrite6": {
-      "page": 22,
-      "row": 256,
+      "page": 36,
+      "row": 343,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 90112
+          "row": 147456
         }
       },
       "epoch": 45,
       "offset": 2048,
-      "documents": 9333548
+      "documents": 9390979
     },
     "openmath": {
-      "page": 2495,
-      "row": 3276,
+      "page": 2511,
+      "row": 1078,
       "start": {
         "file": 23,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 176,
-          "row": 570
+          "row_group": 242,
+          "row": 106
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 10222796
+      "offset": 0,
+      "documents": 10286134
     },
     "ultra_qa": {
-      "page": 1079,
-      "row": 780,
+      "page": 1085,
+      "row": 3490,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 4352
+          "row": 28928
         }
       },
       "epoch": 0,
       "offset": 4096,
-      "documents": 4420364
+      "documents": 4447650
     },
     "ultra_style": {
-      "page": 1647,
-      "row": 173,
+      "page": 1657,
+      "row": 1167,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 6,
-          "row": 72213
+          "row_group": 7,
+          "row": 27375
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 6746285
+      "offset": 4096,
+      "documents": 6788239
     },
     "cortex": {
-      "page": 212069,
-      "row": 58,
+      "page": 213380,
+      "row": 6,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 27144890
+      "documents": 27312646
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 810000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "b860c45edec3c7186db36858be040ec7573c74f1e3787dd00557ac89a4199fc4", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "a24063ea8ec1a680b2cbad4348b11f3720f7b12a48fa11c042f0aa687a24bfbb", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "d3fc1112c30fbcd245d56533cfa5eb69c877b6ac83ebcc2954147694a2f34a08", "coverage.json": "284e9a6216e7c4741fa1f7e01c706d28eb52bfba351b9120d3853f5e4505d59b", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000810000"}
+{"step": 815000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "fdf8df44c98676caf03f6204fff629f95101410723ef61ce23662a551a8f427b", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "125e3addd8b07f3024e8ed1e3c3e2f705e99c3cfcfa6b9f1547db31a2081c525", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "d8c00298782e3f8a28914751297b8d31ce17e96db9eaa5de1147c633462ccb8c", "coverage.json": "1e90d10c39719e43af2b2ea2d77129d1f5aa16ce1b06969bdac4ca8d9ac3d0a6", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000815000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b860c45edec3c7186db36858be040ec7573c74f1e3787dd00557ac89a4199fc4
+oid sha256:fdf8df44c98676caf03f6204fff629f95101410723ef61ce23662a551a8f427b
 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": 76742.70439241717,
-      "required_pages_per_second_upper_estimate": 0.0053092383574745395
+      "required_token_positions_per_second": 76422.60943566434,
+      "required_pages_per_second_upper_estimate": 0.005287093445643752
     },
     {
       "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": 9592.838049052147,
-      "required_pages_per_second_upper_estimate": 0.0040163209699003825
+      "required_token_positions_per_second": 9552.826179458043,
+      "required_pages_per_second_upper_estimate": 0.003999568835644181
     },
     {
       "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": 19185.676098104293,
-      "required_pages_per_second_upper_estimate": 0.0039958703709320606
+      "required_token_positions_per_second": 19105.652358916086,
+      "required_pages_per_second_upper_estimate": 0.003979203536427075
     },
     {
       "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": 28778.514147156442,
-      "required_pages_per_second_upper_estimate": 0.0017252279808752647
+      "required_token_positions_per_second": 28658.47853837413,
+      "required_pages_per_second_upper_estimate": 0.001718032029412527
     },
     {
       "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": 28778.514147156442,
-      "required_pages_per_second_upper_estimate": 0.002237402891635416
+      "required_token_positions_per_second": 28658.47853837413,
+      "required_pages_per_second_upper_estimate": 0.0022280706510334355
     },
     {
       "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": 28778.514147156442,
-      "required_pages_per_second_upper_estimate": 0.3691398795186881
+      "required_token_positions_per_second": 28658.47853837413,
+      "required_pages_per_second_upper_estimate": 0.3676001916134238
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,171 +158,27 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.1648581179296542,
-    "producer_p95_seconds": 0.19809514199914702,
-    "gpu_update_seconds": 0.9607167298014186,
-    "producer_batches_per_second": 6.065822008393333,
-    "gpu_batches_per_second": 1.0408895452530542,
-    "producer_headroom": 5.8275366834611155,
+    "producer_batch_seconds": 0.2541760841795906,
+    "producer_p95_seconds": 0.2388652240006195,
+    "gpu_update_seconds": 0.9647406774570715,
+    "producer_batches_per_second": 3.9342804545428445,
+    "gpu_batches_per_second": 1.0365479795418884,
+    "producer_headroom": 3.795560391021779,
     "sustainable": true,
-    "buffer_seconds": 60.525153977489374,
+    "buffer_seconds": 60.7786626797955,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 81,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 375296.30858700874,
-    "upload_bytes_per_second": 4130574.7776954207,
+    "generation_bytes_per_second": 448277.4506157306,
+    "upload_bytes_per_second": 3886909.442531961,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 80.52985137183853,
+    "estimated_full_shards_per_hour": 96.18990553716601,
     "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": 1.6723634379995929,
-      "bytes": 50251,
-      "pages": 1,
-      "shards": 1
-    },
-    {
-      "seconds": 2.1410375759996896,
-      "bytes": 10006036,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.5198384699997405,
-      "bytes": 2430008,
-      "pages": 30,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2644712469991646,
-      "bytes": 6316580,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.6103071739998995,
-      "bytes": 7727369,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.5575974499997756,
-      "bytes": 11152630,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 1.9014979050007241,
-      "bytes": 5893171,
-      "pages": 27,
-      "shards": 2
-    },
-    {
-      "seconds": 1.375044964999688,
-      "bytes": 2500711,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.54583506400013,
-      "bytes": 5956633,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.6070335669992346,
-      "bytes": 5667674,
-      "pages": 32,
-      "shards": 2
-    },
-    {
-      "seconds": 1.649759572000221,
-      "bytes": 7616382,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.8874110320002728,
-      "bytes": 11916705,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.604149986000266,
-      "bytes": 2375336,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4722619869999107,
-      "bytes": 10022494,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.460268009000174,
-      "bytes": 1130080,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.6416546800001015,
-      "bytes": 7689453,
-      "pages": 38,
-      "shards": 4
-    },
-    {
-      "seconds": 1.5363667999999961,
-      "bytes": 6120972,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.693875957000273,
-      "bytes": 10707952,
-      "pages": 29,
-      "shards": 3
-    },
-    {
-      "seconds": 1.540582020000329,
-      "bytes": 7589375,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.573228470999311,
-      "bytes": 6089737,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.5443801999999778,
-      "bytes": 7496481,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4706297409993567,
-      "bytes": 5704544,
-      "pages": 26,
-      "shards": 2
-    },
-    {
-      "seconds": 1.637408425999638,
-      "bytes": 1096275,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.9842258310000034,
-      "bytes": 13303551,
-      "pages": 43,
-      "shards": 4
-    },
     {
       "seconds": 1.4946416839993617,
       "bytes": 5892242,
@@ -466,6 +322,246 @@
       "bytes": 12131753,
       "pages": 31,
       "shards": 3
+    },
+    {
+      "seconds": 1.6632540899991,
+      "bytes": 5726975,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.7110617899998033,
+      "bytes": 7430334,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.6944532489997073,
+      "bytes": 1160058,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 2.4445176089993765,
+      "bytes": 7341476,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.587468388999696,
+      "bytes": 5869829,
+      "pages": 27,
+      "shards": 2
+    },
+    {
+      "seconds": 1.9299586070010264,
+      "bytes": 10891963,
+      "pages": 40,
+      "shards": 3
+    },
+    {
+      "seconds": 1.62565908400029,
+      "bytes": 7724112,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 1.4196104830007243,
+      "bytes": 988295,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 2.0304724819998228,
+      "bytes": 10476032,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.5808353909997095,
+      "bytes": 7239609,
+      "pages": 35,
+      "shards": 3
+    },
+    {
+      "seconds": 1.859183167000083,
+      "bytes": 1137645,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 4.019235966999986,
+      "bytes": 13232606,
+      "pages": 36,
+      "shards": 4
+    },
+    {
+      "seconds": 1.720672621000631,
+      "bytes": 611156,
+      "pages": 21,
+      "shards": 1
+    },
+    {
+      "seconds": 1.6194256800008588,
+      "bytes": 9820718,
+      "pages": 47,
+      "shards": 3
+    },
+    {
+      "seconds": 1.6360844549999456,
+      "bytes": 7454978,
+      "pages": 29,
+      "shards": 3
+    },
+    {
+      "seconds": 1.568186391999916,
+      "bytes": 1164036,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 2.5616059220010357,
+      "bytes": 11807815,
+      "pages": 42,
+      "shards": 4
+    },
+    {
+      "seconds": 1.6904037299991614,
+      "bytes": 12097177,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2079874939990987,
+      "bytes": 1225243,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.7503577689985832,
+      "bytes": 11722749,
+      "pages": 33,
+      "shards": 4
+    },
+    {
+      "seconds": 1.717468271999678,
+      "bytes": 6201366,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.4180366079999658,
+      "bytes": 818969,
+      "pages": 25,
+      "shards": 1
+    },
+    {
+      "seconds": 1.935665021999739,
+      "bytes": 7829235,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 1.7852341269990575,
+      "bytes": 11660999,
+      "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.8856627909990493,
+      "bytes": 7118698,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.6447518009990745,
+      "bytes": 6129393,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.743864690999544,
+      "bytes": 1054003,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.5393589430004795,
+      "bytes": 11424070,
+      "pages": 35,
+      "shards": 4
+    },
+    {
+      "seconds": 1.9194490070003667,
+      "bytes": 6288888,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.4686280049991183,
+      "bytes": 8470485,
+      "pages": 37,
+      "shards": 4
+    },
+    {
+      "seconds": 1.9930366190001223,
+      "bytes": 5866239,
+      "pages": 26,
+      "shards": 2
+    },
+    {
+      "seconds": 2.011558195001271,
+      "bytes": 5842602,
+      "pages": 41,
+      "shards": 2
+    },
+    {
+      "seconds": 1.6693870859999151,
+      "bytes": 7612861,
+      "pages": 29,
+      "shards": 3
+    },
+    {
+      "seconds": 1.3870315900003334,
+      "bytes": 1261477,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 1.6420892379992438,
+      "bytes": 7607765,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 1.5253431410001212,
+      "bytes": 5688723,
+      "pages": 19,
+      "shards": 2
+    },
+    {
+      "seconds": 1.593651556999248,
+      "bytes": 11614889,
+      "pages": 46,
+      "shards": 3
+    },
+    {
+      "seconds": 1.818591990000641,
+      "bytes": 7461958,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.4790617950002343,
+      "bytes": 1048131,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3689376410002296,
+      "bytes": 5834070,
+      "pages": 28,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a24063ea8ec1a680b2cbad4348b11f3720f7b12a48fa11c042f0aa687a24bfbb
-size 721768959
+oid sha256:125e3addd8b07f3024e8ed1e3c3e2f705e99c3cfcfa6b9f1547db31a2081c525
+size 721763839