Qwen90M step 525000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 520000,
+  "step": 525000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "94689d39f31e52e9199eeddbbd7aaa83322651ef0705b3f99ac0f5afe885fe7f",
+    "model.safetensors": "555a0b375b5d8b6df17309058cb3fac983e079e022e3453c5bb40694dc0dfd3c",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "84af1488eb650b70be0d37c145b2c99c81009be6ddae8e4bb3426d935b6fde18",
+    "trainer.pt": "550740c374917e84e0c390d9c1282f579154291db3c66058d82c3775128b44da",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "234744063e094dfd7b525da4e181ac493fac400c830f871c7e8c59127ed10dc9",
-    "shard_scheduler.json": "ca45b34a5ce482493ede09362d54c536a11ccac97331918b9089c1ba2346d3c8",
-    "coverage.json": "9839d8d5634e978124cc5b5644cf2998b990933ad09e03d5f004fa841278743f",
+    "shard_scheduler.json": "ded67c5cf32a54e2366cba7242e168a5494026aca2d41bea5b359a7947e97553",
+    "coverage.json": "8e20b86d8845516d961760222cf2e262fe8d2ed820d4ef25ec6904b4dc6e9d4b",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 520000,
+  "step": 525000,
   "stats": {
-    "climbmix/blocks": 18720000,
-    "climbmix/targets": 31291918409,
-    "climbmix/input_tokens": 31293901346,
-    "ultra_qa/blocks": 7020000,
-    "ultra_qa/targets": 11452247156,
-    "ultra_qa/input_tokens": 11452269248,
-    "cortex/blocks": 7020000,
-    "cortex/targets": 266976897,
-    "cortex/input_tokens": 11159205229,
-    "ultra_style/blocks": 7020000,
-    "ultra_style/targets": 11102583967,
-    "ultra_style/input_tokens": 11103166058,
-    "openmath/blocks": 4680000,
-    "openmath/targets": 7231495445,
-    "openmath/input_tokens": 7233904958,
-    "rewrite6/blocks": 2340000,
-    "rewrite6/targets": 3836125690,
-    "rewrite6/input_tokens": 3840331312
+    "climbmix/blocks": 18900000,
+    "climbmix/targets": 31592797042,
+    "climbmix/input_tokens": 31594799016,
+    "ultra_qa/blocks": 7087500,
+    "ultra_qa/targets": 11562286610,
+    "ultra_qa/input_tokens": 11562308928,
+    "cortex/blocks": 7087500,
+    "cortex/targets": 269544792,
+    "cortex/input_tokens": 11266328102,
+    "ultra_style/blocks": 7087500,
+    "ultra_style/targets": 11209247634,
+    "ultra_style/input_tokens": 11209834992,
+    "openmath/blocks": 4725000,
+    "openmath/targets": 7301172470,
+    "openmath/input_tokens": 7303605185,
+    "rewrite6/blocks": 2362500,
+    "rewrite6/targets": 3872985034,
+    "rewrite6/input_tokens": 3877231666
   },
   "sources": {
     "climbmix": {
-      "page": 2564,
-      "row": 700,
+      "page": 2588,
+      "row": 3394,
       "start": {
-        "file": 124,
+        "file": 125,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 6,
+          "row_group": 20,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 10502844
+      "documents": 10603842
     },
     "rewrite6": {
-      "page": 8,
-      "row": 2036,
+      "page": 22,
+      "row": 2978,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 32768
+          "row": 90112
         }
       },
       "epoch": 29,
       "offset": 0,
-      "documents": 5991520
+      "documents": 6049806
     },
     "openmath": {
-      "page": 1602,
-      "row": 1247,
+      "page": 1617,
+      "row": 2871,
       "start": {
         "file": 15,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 12,
-          "row": 42
+          "row_group": 73,
+          "row": 482
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 6563039
+      "documents": 6626103
     },
     "ultra_qa": {
-      "page": 692,
-      "row": 3292,
+      "page": 699,
+      "row": 1991,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 42729
+          "row_group": 4,
+          "row": 6884
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 2837724
+      "documents": 2865095
     },
     "ultra_style": {
-      "page": 1057,
-      "row": 1722,
+      "page": 1067,
+      "row": 1860,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 2,
-          "row": 49196
+          "row_group": 3,
+          "row": 4718
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 4331194
+      "documents": 4372292
     },
     "cortex": {
-      "page": 136152,
-      "row": 51,
+      "page": 137458,
+      "row": 115,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 17427507
+      "documents": 17594739
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 520000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "94689d39f31e52e9199eeddbbd7aaa83322651ef0705b3f99ac0f5afe885fe7f", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "84af1488eb650b70be0d37c145b2c99c81009be6ddae8e4bb3426d935b6fde18", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "234744063e094dfd7b525da4e181ac493fac400c830f871c7e8c59127ed10dc9", "shard_scheduler.json": "ca45b34a5ce482493ede09362d54c536a11ccac97331918b9089c1ba2346d3c8", "coverage.json": "9839d8d5634e978124cc5b5644cf2998b990933ad09e03d5f004fa841278743f", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000520000"}
+{"step": 525000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "555a0b375b5d8b6df17309058cb3fac983e079e022e3453c5bb40694dc0dfd3c", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "550740c374917e84e0c390d9c1282f579154291db3c66058d82c3775128b44da", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "234744063e094dfd7b525da4e181ac493fac400c830f871c7e8c59127ed10dc9", "shard_scheduler.json": "ded67c5cf32a54e2366cba7242e168a5494026aca2d41bea5b359a7947e97553", "coverage.json": "8e20b86d8845516d961760222cf2e262fe8d2ed820d4ef25ec6904b4dc6e9d4b", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000525000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:94689d39f31e52e9199eeddbbd7aaa83322651ef0705b3f99ac0f5afe885fe7f
+oid sha256:555a0b375b5d8b6df17309058cb3fac983e079e022e3453c5bb40694dc0dfd3c
 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": 76689.73514938472,
-      "required_pages_per_second_upper_estimate": 0.005305573822335977
+      "required_token_positions_per_second": 77213.6919826638,
+      "required_pages_per_second_upper_estimate": 0.005341822371809575
     },
     {
       "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": 9586.21689367309,
-      "required_pages_per_second_upper_estimate": 0.004013548830408618
+      "required_token_positions_per_second": 9651.711497832976,
+      "required_pages_per_second_upper_estimate": 0.004040970053487503
     },
     {
       "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": 19172.43378734618,
-      "required_pages_per_second_upper_estimate": 0.003993112346824364
+      "required_token_positions_per_second": 19303.42299566595,
+      "required_pages_per_second_upper_estimate": 0.00402039394449965
     },
     {
       "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": 28758.65068101927,
-      "required_pages_per_second_upper_estimate": 0.0017240371964100977
+      "required_token_positions_per_second": 28955.13449349893,
+      "required_pages_per_second_upper_estimate": 0.001735816100954843
     },
     {
       "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": 28758.65068101927,
-      "required_pages_per_second_upper_estimate": 0.002235858594513405
+      "required_token_positions_per_second": 28955.13449349893,
+      "required_pages_per_second_upper_estimate": 0.002251134346692743
     },
     {
       "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": 28758.65068101927,
-      "required_pages_per_second_upper_estimate": 0.3688850923028087
+      "required_token_positions_per_second": 28955.13449349893,
+      "required_pages_per_second_upper_estimate": 0.37140537568141674
     }
   ],
   "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": 704566,
-    "producer_batch_seconds": 0.12374689964809704,
-    "producer_p95_seconds": 0.2969493060008972,
-    "gpu_update_seconds": 0.9613802923739986,
-    "producer_batches_per_second": 8.081010537183005,
-    "gpu_batches_per_second": 1.0401711039141808,
-    "producer_headroom": 7.76892427291436,
+    "producer_batch_seconds": 0.11904607814852852,
+    "producer_p95_seconds": 0.3377903690052335,
+    "gpu_update_seconds": 0.9548565559661825,
+    "producer_batches_per_second": 8.400108727247144,
+    "gpu_batches_per_second": 1.0472777232891683,
+    "producer_headroom": 8.02089888904068,
     "sustainable": true,
-    "buffer_seconds": 60.566958419561914,
+    "buffer_seconds": 60.1559630258695,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 82,
+    "upload_pages_per_commit": 78,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 353411.1862394783,
-    "upload_bytes_per_second": 5844464.712245397,
+    "generation_bytes_per_second": 384588.48339903663,
+    "upload_bytes_per_second": 3837530.023601385,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 75.83381357563269,
+    "estimated_full_shards_per_hour": 82.52373577574086,
     "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": 1.3237921879990608,
-      "bytes": 13505764,
+      "seconds": 1.521548144002736,
+      "bytes": 10001036,
       "pages": 42,
-      "shards": 5
-    },
-    {
-      "seconds": 1.3861956760010798,
-      "bytes": 10089467,
-      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.1900551170001563,
-      "bytes": 2519802,
-      "pages": 34,
+      "seconds": 1.037408388001495,
+      "bytes": 2352962,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.0123512749996735,
-      "bytes": 5925157,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.4666764929934288,
+      "bytes": 11848768,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.84063605999836,
-      "bytes": 1225280,
-      "pages": 36,
-      "shards": 1
+      "seconds": 1.1809732929978054,
+      "bytes": 12035253,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 1.2510575380001683,
-      "bytes": 12042377,
-      "pages": 34,
-      "shards": 4
+      "seconds": 0.976799342999584,
+      "bytes": 895629,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.23156422700049,
-      "bytes": 11911377,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.1010828580037924,
+      "bytes": 6038977,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 0.8838252950008609,
-      "bytes": 2597092,
-      "pages": 37,
+      "seconds": 1.0995818070005043,
+      "bytes": 2583346,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.0776322439996875,
-      "bytes": 6282866,
-      "pages": 37,
+      "seconds": 0.9759389660030138,
+      "bytes": 6031085,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.0567811459986842,
-      "bytes": 4557138,
-      "pages": 29,
+      "seconds": 1.012582071001816,
+      "bytes": 5087118,
+      "pages": 25,
       "shards": 2
     },
     {
-      "seconds": 1.3797404860015376,
-      "bytes": 7238899,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.1673960980042466,
+      "bytes": 13547167,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.0933739200008858,
-      "bytes": 6120770,
-      "pages": 37,
+      "seconds": 1.3716537860018434,
+      "bytes": 6334955,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 0.9898015409999061,
-      "bytes": 867435,
-      "pages": 29,
-      "shards": 1
+      "seconds": 1.127826156996889,
+      "bytes": 2574706,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.4056065479999234,
-      "bytes": 17665783,
+      "seconds": 1.1363400170012028,
+      "bytes": 10297243,
       "pages": 34,
-      "shards": 5
+      "shards": 3
     },
     {
-      "seconds": 1.0921460530007607,
-      "bytes": 6276750,
+      "seconds": 1.2236347540019779,
+      "bytes": 6431007,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0008894900020096,
-      "bytes": 2319506,
-      "pages": 29,
+      "seconds": 0.9796522880060365,
+      "bytes": 2218113,
+      "pages": 23,
       "shards": 2
     },
     {
-      "seconds": 1.162798567002028,
-      "bytes": 6350229,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.3331365960038966,
+      "bytes": 12026403,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.0111840479985403,
-      "bytes": 5239532,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.336267955994117,
+      "bytes": 7188821,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.2082140299971798,
-      "bytes": 13368136,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.3482005179976113,
+      "bytes": 6310167,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.2696354079998855,
-      "bytes": 6070118,
-      "pages": 28,
+      "seconds": 1.184783762000734,
+      "bytes": 5973673,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 0.9898074049997376,
-      "bytes": 2654933,
-      "pages": 38,
+      "seconds": 0.9036034050004673,
+      "bytes": 2269487,
+      "pages": 25,
       "shards": 2
     },
     {
-      "seconds": 1.3211888400001044,
-      "bytes": 9995304,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.6477199560031295,
+      "bytes": 16059670,
+      "pages": 45,
+      "shards": 5
     },
     {
-      "seconds": 0.8846473650010012,
-      "bytes": 1315669,
-      "pages": 36,
-      "shards": 1
+      "seconds": 1.0709953250043327,
+      "bytes": 6125208,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 1.6196549160013092,
-      "bytes": 7467278,
-      "pages": 37,
-      "shards": 3
+      "seconds": 0.8623744039941812,
+      "bytes": 2649650,
+      "pages": 39,
+      "shards": 2
     },
     {
-      "seconds": 1.2983564580026723,
-      "bytes": 12027088,
-      "pages": 34,
+      "seconds": 1.4692237859999295,
+      "bytes": 10546134,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.2329098789996351,
-      "bytes": 6852339,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.1281011890023365,
+      "bytes": 2729862,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.050476451000577,
-      "bytes": 6070361,
-      "pages": 29,
+      "seconds": 2.671323365997523,
+      "bytes": 5596441,
+      "pages": 20,
       "shards": 2
     },
     {
-      "seconds": 1.0647863270023663,
-      "bytes": 1121901,
-      "pages": 35,
+      "seconds": 1.0342092059945571,
+      "bytes": 930671,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.296304608000355,
-      "bytes": 7362354,
-      "pages": 34,
+      "seconds": 1.692445526998199,
+      "bytes": 13078461,
+      "pages": 32,
+      "shards": 4
+    },
+    {
+      "seconds": 1.3595879119966412,
+      "bytes": 10348219,
+      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 1.2885511929998756,
-      "bytes": 10359776,
+      "seconds": 1.0515126419995795,
+      "bytes": 1003171,
+      "pages": 30,
+      "shards": 1
+    },
+    {
+      "seconds": 1.000422951001383,
+      "bytes": 7649877,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.3631092460018408,
-      "bytes": 8446837,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.2835066039988305,
+      "bytes": 10437972,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.1454860439989716,
-      "bytes": 6061904,
+      "seconds": 0.8616337460043724,
+      "bytes": 2430044,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.4194609969999874,
-      "bytes": 9675770,
-      "pages": 36,
+      "seconds": 1.3565980959974695,
+      "bytes": 12199549,
+      "pages": 43,
       "shards": 3
     },
     {
-      "seconds": 1.3136005369960913,
-      "bytes": 2685859,
-      "pages": 36,
+      "seconds": 1.012810416999855,
+      "bytes": 6201884,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3078916630038293,
-      "bytes": 5962936,
-      "pages": 38,
+      "seconds": 0.8832714039963321,
+      "bytes": 2126873,
+      "pages": 22,
       "shards": 2
     },
     {
-      "seconds": 1.0175234589987667,
-      "bytes": 928482,
-      "pages": 27,
-      "shards": 1
-    },
-    {
-      "seconds": 1.5923971939992043,
-      "bytes": 18139508,
+      "seconds": 1.4999150360017666,
+      "bytes": 10166456,
       "pages": 43,
-      "shards": 5
+      "shards": 3
     },
     {
-      "seconds": 0.9787014109970187,
-      "bytes": 5956252,
-      "pages": 28,
+      "seconds": 0.9999620900052832,
+      "bytes": 2535870,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 0.8562311600035173,
-      "bytes": 2647079,
-      "pages": 38,
+      "seconds": 1.0642618760030018,
+      "bytes": 5883611,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.054913633997785,
-      "bytes": 5917684,
-      "pages": 31,
-      "shards": 2
+      "seconds": 2.4567707870010054,
+      "bytes": 16775004,
+      "pages": 44,
+      "shards": 4
     },
     {
-      "seconds": 1.521548144002736,
-      "bytes": 10001036,
-      "pages": 42,
-      "shards": 3
+      "seconds": 1.3269825160023174,
+      "bytes": 2735395,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.037408388001495,
-      "bytes": 2352962,
-      "pages": 29,
+      "seconds": 2.3559383290048572,
+      "bytes": 6286319,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.4666764929934288,
-      "bytes": 11848768,
-      "pages": 38,
-      "shards": 3
+      "seconds": 0.9432465139980195,
+      "bytes": 858963,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.1809732929978054,
-      "bytes": 12035253,
-      "pages": 37,
+      "seconds": 1.6026845509986742,
+      "bytes": 11607409,
+      "pages": 29,
       "shards": 4
     },
     {
-      "seconds": 0.976799342999584,
-      "bytes": 895629,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.04873497600056,
+      "bytes": 6766779,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.1010828580037924,
-      "bytes": 6038977,
+      "seconds": 1.4542191169966827,
+      "bytes": 7891732,
+      "pages": 46,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0971116009968682,
+      "bytes": 6168592,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.0995818070005043,
-      "bytes": 2583346,
+      "seconds": 1.321997156002908,
+      "bytes": 9700555,
       "pages": 34,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 0.9759389660030138,
-      "bytes": 6031085,
-      "pages": 33,
+      "seconds": 1.1995469100002083,
+      "bytes": 2400407,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.012582071001816,
-      "bytes": 5087118,
-      "pages": 25,
+      "seconds": 1.5711508849999518,
+      "bytes": 6147230,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.1673960980042466,
-      "bytes": 13547167,
-      "pages": 41,
-      "shards": 4
+      "seconds": 1.4379745080004795,
+      "bytes": 8148657,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.3716537860018434,
-      "bytes": 6334955,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.2602131539970287,
+      "bytes": 10952650,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.127826156996889,
-      "bytes": 2574706,
-      "pages": 33,
+      "seconds": 1.072258756001247,
+      "bytes": 6000592,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1363400170012028,
-      "bytes": 10297243,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2236347540019779,
-      "bytes": 6431007,
+      "seconds": 0.8994364020036301,
+      "bytes": 2733894,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9796522880060365,
-      "bytes": 2218113,
-      "pages": 23,
+      "seconds": 3.8195672939982614,
+      "bytes": 6121969,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.3331365960038966,
-      "bytes": 12026403,
-      "pages": 40,
+      "seconds": 1.2331268449997879,
+      "bytes": 7295036,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.336267955994117,
-      "bytes": 7188821,
-      "pages": 41,
+      "seconds": 31.867704994001542,
+      "bytes": 12094685,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.3482005179976113,
-      "bytes": 6310167,
-      "pages": 32,
+      "seconds": 1.475563729996793,
+      "bytes": 6268574,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.184783762000734,
-      "bytes": 5973673,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.4652358960011043,
+      "bytes": 7414584,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 0.9036034050004673,
-      "bytes": 2269487,
-      "pages": 25,
+      "seconds": 2.4988657550056814,
+      "bytes": 5294252,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.6477199560031295,
-      "bytes": 16059670,
-      "pages": 45,
-      "shards": 5
+      "seconds": 1.1802646279975306,
+      "bytes": 7575689,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.0709953250043327,
-      "bytes": 6125208,
-      "pages": 31,
-      "shards": 2
+      "seconds": 0.8871877959973062,
+      "bytes": 955502,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 0.8623744039941812,
-      "bytes": 2649650,
+      "seconds": 1.3968777099944418,
+      "bytes": 12101140,
       "pages": 39,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.4692237859999295,
-      "bytes": 10546134,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.1570743730044342,
+      "bytes": 12058095,
+      "pages": 40,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:84af1488eb650b70be0d37c145b2c99c81009be6ddae8e4bb3426d935b6fde18
-size 721776895
+oid sha256:550740c374917e84e0c390d9c1282f579154291db3c66058d82c3775128b44da
+size 721783935