Qwen90M step 40000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 35000,
+  "step": 40000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "257552ddb070147bf74ac81d1972455eac1b5b94298e7820362135ec7d22bf73",
+    "model.safetensors": "1f54edcf173e14eaa220773d6904ca3660dca9333bec4cca80621f41c11ae4ba",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "bdb02210299e7645fb1fcfe319164a683c0fab01d39f95c7ebced4aab5172f8e",
+    "trainer.pt": "ddb86ed5b8deb7d4e788e14749c412fa48be7f3cefd33dadef82d54b10488b02",
     "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2",
     "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e",
-    "shard_scheduler.json": "5414324baf2ef485c06d3f1e0e9a90e7ae02cc5270937d0e6356509393fcac57",
-    "coverage.json": "1384d2eab2585bf506f3fd13f050eebeb4ea9493ce600a26e3511d35382fe101",
+    "shard_scheduler.json": "dd35db67300912a92de8d12b39d40858b1385a9f601def538c82db10aecb48b5",
+    "coverage.json": "496553c7362fb891e4a087907a1ee7bc57bd29676eca9ccbf872f730f0c086c9",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 35000,
+  "step": 40000,
   "stats": {
-    "climbmix/blocks": 1260000,
-    "climbmix/targets": 2105041906,
-    "climbmix/input_tokens": 2105175390,
-    "ultra_qa/blocks": 472500,
-    "ultra_qa/targets": 770911552,
-    "ultra_qa/input_tokens": 770913097,
-    "cortex/blocks": 472500,
-    "cortex/targets": 17983452,
-    "cortex/input_tokens": 751155694,
-    "ultra_style/blocks": 472500,
-    "ultra_style/targets": 746810228,
-    "ultra_style/input_tokens": 746849525,
-    "openmath/blocks": 315000,
-    "openmath/targets": 486799702,
-    "openmath/input_tokens": 486962255,
-    "rewrite6/blocks": 157500,
-    "rewrite6/targets": 258227030,
-    "rewrite6/input_tokens": 258510170
+    "climbmix/blocks": 1440000,
+    "climbmix/targets": 2405645996,
+    "climbmix/input_tokens": 2405798466,
+    "ultra_qa/blocks": 540000,
+    "ultra_qa/targets": 881103422,
+    "ultra_qa/input_tokens": 881105161,
+    "cortex/blocks": 540000,
+    "cortex/targets": 20545868,
+    "cortex/input_tokens": 858494555,
+    "ultra_style/blocks": 540000,
+    "ultra_style/targets": 853948670,
+    "ultra_style/input_tokens": 853993129,
+    "openmath/blocks": 360000,
+    "openmath/targets": 556374754,
+    "openmath/input_tokens": 556560449,
+    "rewrite6/blocks": 180000,
+    "rewrite6/targets": 295083056,
+    "rewrite6/input_tokens": 295406501
   },
   "sources": {
     "climbmix": {
-      "page": 172,
-      "row": 2609,
+      "page": 197,
+      "row": 1098,
       "start": {
-        "file": 8,
+        "file": 9,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 25,
+          "row_group": 43,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 8192,
-      "documents": 707121
+      "offset": 4096,
+      "documents": 808010
     },
     "rewrite6": {
-      "page": 48,
-      "row": 1325,
+      "page": 12,
+      "row": 807,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 196608
+          "row": 49152
         }
       },
-      "epoch": 1,
+      "epoch": 2,
       "offset": 0,
-      "documents": 403337
+      "documents": 460767
     },
     "openmath": {
-      "page": 107,
-      "row": 3947,
+      "page": 123,
+      "row": 1391,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 1,
-          "row": 622
+          "row_group": 67,
+          "row": 158
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 442219
+      "offset": 2048,
+      "documents": 505199
     },
     "ultra_qa": {
-      "page": 46,
-      "row": 2796,
+      "page": 53,
+      "row": 1357,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 2,
-          "row": 59140
+          "row_group": 3,
+          "row": 23181
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 191212
+      "documents": 218445
     },
     "ultra_style": {
-      "page": 71,
-      "row": 1500,
+      "page": 81,
+      "row": 1172,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 3,
-          "row": 35792
+          "row": 76752
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 292316
+      "offset": 0,
+      "documents": 332948
     },
     "cortex": {
-      "page": 9162,
-      "row": 44,
+      "page": 10471,
+      "row": 124,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 1172780
+      "documents": 1340412
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 35000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "257552ddb070147bf74ac81d1972455eac1b5b94298e7820362135ec7d22bf73", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "bdb02210299e7645fb1fcfe319164a683c0fab01d39f95c7ebced4aab5172f8e", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "5414324baf2ef485c06d3f1e0e9a90e7ae02cc5270937d0e6356509393fcac57", "coverage.json": "1384d2eab2585bf506f3fd13f050eebeb4ea9493ce600a26e3511d35382fe101", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000035000"}
+{"step": 40000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "1f54edcf173e14eaa220773d6904ca3660dca9333bec4cca80621f41c11ae4ba", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "ddb86ed5b8deb7d4e788e14749c412fa48be7f3cefd33dadef82d54b10488b02", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "dd35db67300912a92de8d12b39d40858b1385a9f601def538c82db10aecb48b5", "coverage.json": "496553c7362fb891e4a087907a1ee7bc57bd29676eca9ccbf872f730f0c086c9", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000040000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:257552ddb070147bf74ac81d1972455eac1b5b94298e7820362135ec7d22bf73
+oid sha256:1f54edcf173e14eaa220773d6904ca3660dca9333bec4cca80621f41c11ae4ba
 size 360622744

shard_scheduler.json

@@ -22,8 +22,8 @@
       "tokens_per_second": 1062967.6168812835,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 75370.88600524074,
-      "required_pages_per_second_upper_estimate": 0.006532957373170129
+      "required_token_positions_per_second": 75414.82490767069,
+      "required_pages_per_second_upper_estimate": 0.006536765885870627
     },
     {
       "source": "rewrite6",
@@ -36,8 +36,8 @@
       "tokens_per_second": 1206672.263377472,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 9421.360750655092,
-      "required_pages_per_second_upper_estimate": 0.003758321665332333
+      "required_token_positions_per_second": 9426.853113458836,
+      "required_pages_per_second_upper_estimate": 0.003760512650972888
     },
     {
       "source": "openmath",
@@ -50,8 +50,8 @@
       "tokens_per_second": 410042.7581134458,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 18842.721501310185,
-      "required_pages_per_second_upper_estimate": 0.004358931998755937
+      "required_token_positions_per_second": 18853.70622691767,
+      "required_pages_per_second_upper_estimate": 0.004361473121700661
     },
     {
       "source": "ultra_qa",
@@ -64,8 +64,8 @@
       "tokens_per_second": 2163486.5624068812,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28264.082251965276,
-      "required_pages_per_second_upper_estimate": 0.0017211343189655782
+      "required_token_positions_per_second": 28280.55934037651,
+      "required_pages_per_second_upper_estimate": 0.0017221376872012205
     },
     {
       "source": "ultra_style",
@@ -78,8 +78,8 @@
       "tokens_per_second": 1546694.2737633225,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28264.082251965276,
-      "required_pages_per_second_upper_estimate": 0.0021758472917435424
+      "required_token_positions_per_second": 28280.55934037651,
+      "required_pages_per_second_upper_estimate": 0.002177115743620939
     },
     {
       "source": "cortex",
@@ -92,8 +92,8 @@
       "tokens_per_second": 1015177.8395898147,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 22,
-      "required_token_positions_per_second": 28264.082251965276,
-      "required_pages_per_second_upper_estimate": 0.35094530777115207
+      "required_token_positions_per_second": 28280.55934037651,
+      "required_pages_per_second_upper_estimate": 0.3511498980639521
     }
   ],
   "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": 688946,
-    "producer_batch_seconds": 0.1257512091639228,
-    "producer_p95_seconds": 0.3557953109993832,
-    "gpu_update_seconds": 0.9782026443854395,
-    "producer_batches_per_second": 7.952209816896882,
-    "gpu_batches_per_second": 1.0222830675624015,
-    "producer_headroom": 7.778872671596381,
+    "producer_batch_seconds": 0.1152253453594767,
+    "producer_p95_seconds": 0.4034826810020604,
+    "gpu_update_seconds": 0.9776327146587446,
+    "producer_batches_per_second": 8.678646150985522,
+    "gpu_batches_per_second": 1.0228790270680161,
+    "producer_headroom": 8.48452839615064,
     "sustainable": true,
-    "buffer_seconds": 60.648563951897245,
+    "buffer_seconds": 60.61322830884216,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 84,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 387927.91475817043,
-    "upload_bytes_per_second": 4221895.326348876,
+    "generation_bytes_per_second": 346780.9633111328,
+    "upload_bytes_per_second": 3718135.8829697864,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 83.24030000742755,
+    "estimated_full_shards_per_hour": 74.41112207890022,
     "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": 0.8321514729977935,
-      "bytes": 6044525,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4008962299994892,
-      "bytes": 7557731,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 0.8883114669988572,
-      "bytes": 823542,
+      "seconds": 0.9655926410050597,
+      "bytes": 5648372,
       "pages": 25,
-      "shards": 1
-    },
-    {
-      "seconds": 1.2165515029992093,
-      "bytes": 11904873,
-      "pages": 42,
-      "shards": 4
+      "shards": 2
     },
     {
-      "seconds": 0.9436809830003767,
-      "bytes": 889392,
-      "pages": 26,
-      "shards": 1
+      "seconds": 2.317049978002615,
+      "bytes": 7094864,
+      "pages": 41,
+      "shards": 2
     },
     {
-      "seconds": 1.1113080279974383,
-      "bytes": 11835067,
+      "seconds": 1.0156349210010376,
+      "bytes": 12015881,
       "pages": 35,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 0.9863674269981857,
-      "bytes": 7525026,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.0830463239981327,
+      "bytes": 6195752,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.7792704010025773,
-      "bytes": 5012728,
-      "pages": 27,
+      "seconds": 0.7641328610043274,
+      "bytes": 2288843,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0204439990011451,
-      "bytes": 7601581,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.014886224998918,
+      "bytes": 6118927,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.844439466996846,
-      "bytes": 1090867,
-      "pages": 34,
+      "seconds": 0.9393350060054217,
+      "bytes": 807847,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 0.8992618750016845,
-      "bytes": 5707571,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.4682032510027057,
+      "bytes": 16949786,
+      "pages": 39,
+      "shards": 5
     },
     {
-      "seconds": 1.1450656299966795,
-      "bytes": 17577071,
+      "seconds": 0.9623997290036641,
+      "bytes": 6092433,
       "pages": 37,
-      "shards": 5
+      "shards": 2
     },
     {
-      "seconds": 1.0943805140013865,
-      "bytes": 891607,
-      "pages": 26,
+      "seconds": 0.7187483379966579,
+      "bytes": 1132933,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 0.9035562749995734,
-      "bytes": 6454400,
+      "seconds": 1.4302320659990073,
+      "bytes": 11751486,
       "pages": 35,
-      "shards": 2
+      "shards": 4
     },
     {
-      "seconds": 1.2266940429981332,
-      "bytes": 2571680,
-      "pages": 38,
-      "shards": 2
+      "seconds": 0.8936330840006121,
+      "bytes": 843076,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.0315104050023365,
-      "bytes": 10269533,
-      "pages": 32,
+      "seconds": 1.0585707080026623,
+      "bytes": 7490877,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 0.8940106660011224,
-      "bytes": 5798800,
-      "pages": 31,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0886239710016525,
-      "bytes": 8189805,
-      "pages": 33,
+      "seconds": 33.798858237001696,
+      "bytes": 12238534,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.7768403989975923,
-      "bytes": 6370933,
-      "pages": 36,
+      "seconds": 0.911944024002878,
+      "bytes": 5136667,
+      "pages": 41,
       "shards": 2
     },
     {
-      "seconds": 1.1446712159995514,
-      "bytes": 7008906,
-      "pages": 33,
+      "seconds": 1.2792566019998048,
+      "bytes": 7551543,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9077771720003511,
-      "bytes": 6012024,
-      "pages": 36,
+      "seconds": 0.9091981460005627,
+      "bytes": 5814772,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.4875493239997013,
-      "bytes": 6032602,
-      "pages": 34,
+      "seconds": 1.0604712630010908,
+      "bytes": 6743320,
+      "pages": 36,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0128854309950839,
+      "bytes": 6059900,
+      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 0.8995478449978691,
-      "bytes": 2436827,
-      "pages": 30,
+      "seconds": 1.0482589550010744,
+      "bytes": 7056720,
+      "pages": 42,
       "shards": 2
     },
     {
-      "seconds": 1.0387542439966637,
-      "bytes": 11801776,
-      "pages": 24,
+      "seconds": 1.9086563699966064,
+      "bytes": 7678172,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.245932874000573,
-      "bytes": 7158906,
-      "pages": 42,
+      "seconds": 1.1078665649984032,
+      "bytes": 9761910,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.8130094469997857,
-      "bytes": 5715102,
-      "pages": 28,
+      "seconds": 0.90664235100121,
+      "bytes": 2695308,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8107168300011836,
-      "bytes": 6329429,
-      "pages": 36,
+      "seconds": 0.7943546639944543,
+      "bytes": 258770,
+      "pages": 8,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1189483790003578,
+      "bytes": 5921688,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.009647611001128,
-      "bytes": 2308548,
-      "pages": 27,
+      "seconds": 1.2605290909996256,
+      "bytes": 6824656,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 0.98430517000088,
-      "bytes": 10007399,
-      "pages": 44,
+      "seconds": 11.391802897000161,
+      "bytes": 12565360,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 0.9780213199992431,
-      "bytes": 6711146,
-      "pages": 29,
+      "seconds": 0.941881225000543,
+      "bytes": 6023516,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9533051110011002,
-      "bytes": 7632407,
-      "pages": 38,
-      "shards": 3
+      "seconds": 0.9670609070017235,
+      "bytes": 2689085,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.273418650001986,
-      "bytes": 11058609,
+      "seconds": 1.1650014270053362,
+      "bytes": 5886860,
       "pages": 31,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1598776869941503,
+      "bytes": 10363400,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.9772542809987499,
-      "bytes": 2642702,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.1336681269967812,
+      "bytes": 8324153,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 0.8478547829981835,
-      "bytes": 5863488,
-      "pages": 28,
+      "seconds": 1.0718341040046653,
+      "bytes": 6483305,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.6997049530000368,
-      "bytes": 1148192,
-      "pages": 35,
+      "seconds": 0.9573828110005707,
+      "bytes": 873310,
+      "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.378931018000003,
-      "bytes": 13331524,
-      "pages": 36,
+      "seconds": 1.2830225369980326,
+      "bytes": 12398676,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 1.0631361519990605,
-      "bytes": 9774660,
+      "seconds": 1.2898951360039064,
+      "bytes": 6161152,
       "pages": 36,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.0518995679994987,
-      "bytes": 1242140,
-      "pages": 33,
-      "shards": 1
+      "seconds": 0.9979428110018489,
+      "bytes": 2470155,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.0469969259975187,
-      "bytes": 7461003,
+      "seconds": 1.098158920998685,
+      "bytes": 11654949,
       "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.1135563850002654,
-      "bytes": 5487788,
-      "pages": 31,
+      "seconds": 1.0713990720032598,
+      "bytes": 5874732,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0679467740010296,
-      "bytes": 7651397,
-      "pages": 39,
+      "seconds": 0.9352636539988453,
+      "bytes": 6076485,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 0.9655926410050597,
-      "bytes": 5648372,
-      "pages": 25,
+      "seconds": 1.2692730029957602,
+      "bytes": 6086693,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 2.317049978002615,
-      "bytes": 7094864,
-      "pages": 41,
+      "seconds": 1.036868928000331,
+      "bytes": 2682212,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.0156349210010376,
-      "bytes": 12015881,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.1587374009977793,
+      "bytes": 10973458,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.0830463239981327,
-      "bytes": 6195752,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8646178900016821,
+      "bytes": 947338,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 0.7641328610043274,
-      "bytes": 2288843,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.0228583920033998,
+      "bytes": 13452885,
+      "pages": 43,
+      "shards": 4
     },
     {
-      "seconds": 1.014886224998918,
-      "bytes": 6118927,
-      "pages": 37,
+      "seconds": 1.0817572929954622,
+      "bytes": 6130820,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9393350060054217,
-      "bytes": 807847,
-      "pages": 26,
+      "seconds": 0.9214774369975203,
+      "bytes": 1243052,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.4682032510027057,
-      "bytes": 16949786,
-      "pages": 39,
-      "shards": 5
+      "seconds": 1.3106369909946807,
+      "bytes": 11452061,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 0.9623997290036641,
-      "bytes": 6092433,
+      "seconds": 1.0961634519990184,
+      "bytes": 6066738,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.7187483379966579,
-      "bytes": 1132933,
-      "pages": 34,
-      "shards": 1
+      "seconds": 0.9469924939985503,
+      "bytes": 890694,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.4302320659990073,
-      "bytes": 11751486,
-      "pages": 35,
+      "seconds": 1.2476149669964798,
+      "bytes": 13021486,
+      "pages": 30,
       "shards": 4
     },
     {
-      "seconds": 0.8936330840006121,
-      "bytes": 843076,
-      "pages": 27,
-      "shards": 1
+      "seconds": 0.8867707609970239,
+      "bytes": 6143048,
+      "pages": 43,
+      "shards": 2
     },
     {
-      "seconds": 1.0585707080026623,
-      "bytes": 7490877,
-      "pages": 34,
+      "seconds": 1.05023945299763,
+      "bytes": 7400872,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 33.798858237001696,
-      "bytes": 12238534,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.1047442850031075,
+      "bytes": 6203278,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.911944024002878,
-      "bytes": 5136667,
-      "pages": 41,
-      "shards": 2
+      "seconds": 1.0125872680000612,
+      "bytes": 1072664,
+      "pages": 31,
+      "shards": 1
     },
     {
-      "seconds": 1.2792566019998048,
-      "bytes": 7551543,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.132626735001395,
+      "bytes": 11459123,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 0.9091981460005627,
-      "bytes": 5814772,
-      "pages": 27,
+      "seconds": 1.060930259001907,
+      "bytes": 6808205,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.0604712630010908,
-      "bytes": 6743320,
-      "pages": 36,
-      "shards": 4
+      "seconds": 3.3303450339953997,
+      "bytes": 5952538,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.0128854309950839,
-      "bytes": 6059900,
-      "pages": 26,
-      "shards": 2
+      "seconds": 0.9529956519982079,
+      "bytes": 7196338,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.0482589550010744,
-      "bytes": 7056720,
-      "pages": 42,
+      "seconds": 1.1464728400023887,
+      "bytes": 6003121,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.9086563699966064,
-      "bytes": 7678172,
-      "pages": 33,
+      "seconds": 0.9798986519963364,
+      "bytes": 7781027,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.1078665649984032,
-      "bytes": 9761910,
-      "pages": 33,
+      "seconds": 0.9489700769991032,
+      "bytes": 911408,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3251626939963899,
+      "bytes": 11870118,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 0.90664235100121,
-      "bytes": 2695308,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.1823667450007633,
+      "bytes": 10876587,
+      "pages": 32,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bdb02210299e7645fb1fcfe319164a683c0fab01d39f95c7ebced4aab5172f8e
-size 721769407
+oid sha256:ddb86ed5b8deb7d4e788e14749c412fa48be7f3cefd33dadef82d54b10488b02
+size 721776959