Qwen90M step 55000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 50000,
+  "step": 55000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "681b4da7833f165cc654739e94f0e7a807bd7a54f8adefbfe0a179fa5fd48b20",
+    "model.safetensors": "f37360b73b215cfbd1cab3608606dbff58530c087eebca9f4fe5d5a1c20784da",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "ad5414aa9cc5e09a287ca937f00d1a790494a4a5870ec4a0ca71e8a2bf9bfdc0",
+    "trainer.pt": "54b03f408ed31fec8a13f34a9c659787ebd6b16e7fa218e2cea4dd5412383f2f",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f",
-    "shard_scheduler.json": "7b240c248709a7609872c5ec8f72f1fa0583a80ceadb696ffbc883b68d067333",
-    "coverage.json": "5e6ec826d3400c038f929348e9886127bd7b7fc2f4b301d6ad5272d1b42e325c",
+    "shard_scheduler.json": "0b7760fce951fcc476bd9ce864c73c4a8111a03bf92e84c8373e26e818a1f854",
+    "coverage.json": "a90e5fa0fd4f8822ac17717d8ddbef2645fe20188ab34ad850e1754843896bc9",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 50000,
+  "step": 55000,
   "stats": {
-    "climbmix/blocks": 1800000,
-    "climbmix/targets": 3007990030,
-    "climbmix/input_tokens": 3008180540,
-    "ultra_qa/blocks": 675000,
-    "ultra_qa/targets": 1101475594,
-    "ultra_qa/input_tokens": 1101477776,
-    "cortex/blocks": 675000,
-    "cortex/targets": 25668963,
-    "cortex/input_tokens": 1073030847,
-    "ultra_style/blocks": 675000,
-    "ultra_style/targets": 1067349516,
-    "ultra_style/input_tokens": 1067405409,
-    "openmath/blocks": 450000,
-    "openmath/targets": 695410628,
-    "openmath/input_tokens": 695642540,
-    "rewrite6/blocks": 225000,
-    "rewrite6/targets": 368835030,
-    "rewrite6/input_tokens": 369239542
+    "climbmix/blocks": 1980000,
+    "climbmix/targets": 3309029909,
+    "climbmix/input_tokens": 3309239389,
+    "ultra_qa/blocks": 742500,
+    "ultra_qa/targets": 1211473004,
+    "ultra_qa/input_tokens": 1211475401,
+    "cortex/blocks": 742500,
+    "cortex/targets": 28237480,
+    "cortex/input_tokens": 1180342911,
+    "ultra_style/blocks": 742500,
+    "ultra_style/targets": 1174419713,
+    "ultra_style/input_tokens": 1174481204,
+    "openmath/blocks": 495000,
+    "openmath/targets": 764845062,
+    "openmath/input_tokens": 765100091,
+    "rewrite6/blocks": 247500,
+    "rewrite6/targets": 405729377,
+    "rewrite6/input_tokens": 406174162
   },
   "sources": {
     "climbmix": {
-      "page": 246,
-      "row": 1907,
+      "page": 271,
+      "row": 208,
       "start": {
-        "file": 11,
+        "file": 13,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 73,
+          "row_group": 9,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 108544,
-      "documents": 1009523
+      "offset": 53248,
+      "documents": 1110224
     },
     "rewrite6": {
-      "page": 40,
-      "row": 1649,
+      "page": 4,
+      "row": 1053,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 163840
+          "row": 16384
         }
       },
-      "epoch": 2,
-      "offset": 0,
-      "documents": 576297
+      "epoch": 3,
+      "offset": 2048,
+      "documents": 633649
     },
     "openmath": {
-      "page": 154,
-      "row": 365,
+      "page": 169,
+      "row": 1982,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 194,
-          "row": 134
+          "row_group": 255,
+          "row": 574
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 631149
+      "offset": 2048,
+      "documents": 694206
     },
     "ultra_qa": {
-      "page": 66,
-      "row": 2711,
+      "page": 73,
+      "row": 1336,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 12002
+          "row": 40674
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 273047
+      "offset": 4096,
+      "documents": 300344
     },
     "ultra_style": {
-      "page": 101,
-      "row": 3047,
+      "page": 111,
+      "row": 3394,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 4,
-          "row": 73675
+          "row_group": 5,
+          "row": 28852
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 416743
+      "offset": 2048,
+      "documents": 458050
     },
     "cortex": {
-      "page": 13088,
-      "row": 13,
+      "page": 14397,
+      "row": 119,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 1675277
+      "documents": 1842935
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 50000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "681b4da7833f165cc654739e94f0e7a807bd7a54f8adefbfe0a179fa5fd48b20", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "ad5414aa9cc5e09a287ca937f00d1a790494a4a5870ec4a0ca71e8a2bf9bfdc0", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f", "shard_scheduler.json": "7b240c248709a7609872c5ec8f72f1fa0583a80ceadb696ffbc883b68d067333", "coverage.json": "5e6ec826d3400c038f929348e9886127bd7b7fc2f4b301d6ad5272d1b42e325c", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000050000"}
+{"step": 55000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "f37360b73b215cfbd1cab3608606dbff58530c087eebca9f4fe5d5a1c20784da", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "54b03f408ed31fec8a13f34a9c659787ebd6b16e7fa218e2cea4dd5412383f2f", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f", "shard_scheduler.json": "0b7760fce951fcc476bd9ce864c73c4a8111a03bf92e84c8373e26e818a1f854", "coverage.json": "a90e5fa0fd4f8822ac17717d8ddbef2645fe20188ab34ad850e1754843896bc9", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000055000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:681b4da7833f165cc654739e94f0e7a807bd7a54f8adefbfe0a179fa5fd48b20
+oid sha256:f37360b73b215cfbd1cab3608606dbff58530c087eebca9f4fe5d5a1c20784da
 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": 76046.63191128224,
-      "required_pages_per_second_upper_estimate": 0.006591529315643466
+      "required_token_positions_per_second": 76127.77658254562,
+      "required_pages_per_second_upper_estimate": 0.006598562730089286
     },
     {
       "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": 9505.82898891028,
-      "required_pages_per_second_upper_estimate": 0.0037920173084850326
+      "required_token_positions_per_second": 9515.972072818202,
+      "required_pages_per_second_upper_estimate": 0.0037960635363085216
     },
     {
       "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": 19011.65797782056,
-      "required_pages_per_second_upper_estimate": 0.004398012479416172
+      "required_token_positions_per_second": 19031.944145636404,
+      "required_pages_per_second_upper_estimate": 0.004402705327316008
     },
     {
       "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": 28517.48696673084,
-      "required_pages_per_second_upper_estimate": 0.0017365653365830128
+      "required_token_positions_per_second": 28547.916218454604,
+      "required_pages_per_second_upper_estimate": 0.0017384183183630445
     },
     {
       "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": 28517.48696673084,
-      "required_pages_per_second_upper_estimate": 0.002195355088155342
+      "required_token_positions_per_second": 28547.916218454604,
+      "required_pages_per_second_upper_estimate": 0.002197697616186597
     },
     {
       "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": 28517.48696673084,
-      "required_pages_per_second_upper_estimate": 0.3540917462375162
+      "required_token_positions_per_second": 28547.916218454604,
+      "required_pages_per_second_upper_estimate": 0.35446957570377097
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,386 +158,410 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.07752646051569911,
-    "producer_p95_seconds": 0.21125603599830356,
-    "gpu_update_seconds": 0.9695103931231667,
-    "producer_batches_per_second": 12.898821813198861,
-    "gpu_batches_per_second": 1.0314484579980774,
-    "producer_headroom": 12.505541806940107,
+    "producer_batch_seconds": 0.09582121260153542,
+    "producer_p95_seconds": 0.27117204399837647,
+    "gpu_update_seconds": 0.9684769910501257,
+    "producer_batches_per_second": 10.436102537737831,
+    "gpu_batches_per_second": 1.0325490530401695,
+    "producer_headroom": 10.107125184038916,
     "sustainable": true,
-    "buffer_seconds": 60.109644373636336,
+    "buffer_seconds": 60.0455734451078,
     "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": 376788.25974089396,
-    "upload_bytes_per_second": 6017077.810174047,
+    "generation_bytes_per_second": 416734.52252620424,
+    "upload_bytes_per_second": 6385147.086400999,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 80.84998935861697,
+    "estimated_full_shards_per_hour": 89.42152745093914,
     "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": 1.0632401930001834,
-      "bytes": 22792,
-      "pages": 1,
-      "shards": 1
+      "seconds": 1.8200730230000772,
+      "bytes": 7217811,
+      "pages": 25,
+      "shards": 3
     },
     {
-      "seconds": 0.9811290180000469,
-      "bytes": 1073130,
-      "pages": 29,
+      "seconds": 0.9584536009997464,
+      "bytes": 5627627,
+      "pages": 33,
+      "shards": 2
+    },
+    {
+      "seconds": 1.156600534000063,
+      "bytes": 13506264,
+      "pages": 39,
+      "shards": 4
+    },
+    {
+      "seconds": 0.817877588000556,
+      "bytes": 6663126,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9133121940003548,
+      "bytes": 1030486,
+      "pages": 31,
       "shards": 1
     },
     {
-      "seconds": 1.3945427510002446,
-      "bytes": 16977562,
-      "pages": 43,
-      "shards": 5
+      "seconds": 1.181683423999857,
+      "bytes": 12141179,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 0.7539118750000853,
-      "bytes": 789098,
-      "pages": 25,
+      "seconds": 1.1883426349995716,
+      "bytes": 1212298,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 0.9344837560001906,
-      "bytes": 6249982,
+      "seconds": 1.1447565099988424,
+      "bytes": 5933026,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.0289287979999244,
-      "bytes": 7839831,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.2180850740005553,
+      "bytes": 13486469,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.02437976300007,
-      "bytes": 6057442,
-      "pages": 31,
+      "seconds": 1.0477722719988378,
+      "bytes": 4809637,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1024301279999236,
-      "bytes": 7712415,
+      "seconds": 1.1933620469990274,
+      "bytes": 7601419,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.9960975080002754,
-      "bytes": 6877408,
-      "pages": 35,
+      "seconds": 1.0362921949999873,
+      "bytes": 5998979,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.0195401410001068,
-      "bytes": 6014632,
-      "pages": 34,
-      "shards": 2
+      "seconds": 0.7888650649983902,
+      "bytes": 730238,
+      "pages": 23,
+      "shards": 1
     },
     {
-      "seconds": 1.221793560000151,
-      "bytes": 10970126,
-      "pages": 34,
+      "seconds": 1.1390384229998745,
+      "bytes": 12216394,
+      "pages": 44,
       "shards": 4
     },
     {
-      "seconds": 0.999679897000533,
-      "bytes": 1109612,
-      "pages": 33,
-      "shards": 1
+      "seconds": 0.9729804899998271,
+      "bytes": 6782943,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 0.9920720150003035,
-      "bytes": 6122911,
-      "pages": 35,
+      "seconds": 1.0282262380005704,
+      "bytes": 6456198,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0328473609997673,
-      "bytes": 7480746,
-      "pages": 29,
+      "seconds": 1.007465778999176,
+      "bytes": 7569670,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.141071416000159,
-      "bytes": 10868715,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.248841076001554,
+      "bytes": 5066284,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.1746170310007074,
-      "bytes": 7476573,
+      "seconds": 1.0381730779990903,
+      "bytes": 7573812,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0039610859994355,
-      "bytes": 1092998,
-      "pages": 34,
+      "seconds": 0.756838974000857,
+      "bytes": 1060731,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.1609742079999705,
-      "bytes": 10556650,
-      "pages": 33,
+      "seconds": 1.1045795680001902,
+      "bytes": 11876139,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.1113881370001764,
-      "bytes": 7415466,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.0963092559995857,
+      "bytes": 12324311,
+      "pages": 39,
+      "shards": 5
     },
     {
-      "seconds": 1.0273279060002096,
-      "bytes": 1056303,
-      "pages": 33,
+      "seconds": 0.7185119169989775,
+      "bytes": 873551,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 1.1538401379993957,
-      "bytes": 11570054,
-      "pages": 22,
+      "seconds": 0.8877478260001226,
+      "bytes": 6165145,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9784168599999248,
+      "bytes": 2018917,
+      "pages": 19,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1833946610004205,
+      "bytes": 10020680,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 0.7532719520004321,
-      "bytes": 2586371,
+      "seconds": 0.9556004539990681,
+      "bytes": 6956463,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9901489570002013,
-      "bytes": 9709695,
-      "pages": 35,
+      "seconds": 1.286864922998575,
+      "bytes": 7382695,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 0.8841591720001816,
-      "bytes": 855761,
-      "pages": 25,
-      "shards": 1
+      "seconds": 0.9901690320002672,
+      "bytes": 6106536,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.0832384449995516,
-      "bytes": 7778060,
-      "pages": 38,
+      "seconds": 1.1473440179997851,
+      "bytes": 6606039,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.9103183699999136,
-      "bytes": 5940010,
+      "seconds": 1.092060394001237,
+      "bytes": 6397904,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9267724249993989,
-      "bytes": 5041964,
-      "pages": 22,
-      "shards": 2
+      "seconds": 0.9357377939995786,
+      "bytes": 1197833,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 1.078922188999968,
-      "bytes": 13377142,
-      "pages": 44,
-      "shards": 5
+      "seconds": 1.0255696429994714,
+      "bytes": 13323829,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 0.9435872209996887,
-      "bytes": 6172084,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.3980831969984138,
+      "bytes": 10844649,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 0.8994586850003543,
-      "bytes": 2397832,
+      "seconds": 0.9555492769995908,
+      "bytes": 915953,
       "pages": 28,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 1.344766387999698,
-      "bytes": 10673063,
+      "seconds": 1.0586995369994838,
+      "bytes": 7742624,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 2.8867894830000296,
-      "bytes": 1267136,
+      "seconds": 0.8267605399996683,
+      "bytes": 1098322,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.532430633000331,
-      "bytes": 13601194,
-      "pages": 37,
-      "shards": 4
-    },
-    {
-      "seconds": 1.1585651320001489,
-      "bytes": 6030177,
-      "pages": 31,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9087623919995167,
-      "bytes": 6312525,
-      "pages": 37,
+      "seconds": 1.0487060740015295,
+      "bytes": 7770470,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.0277426739994553,
-      "bytes": 6105617,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.5646253450013319,
+      "bytes": 15407220,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 0.9305132150002464,
-      "bytes": 1095891,
-      "pages": 34,
+      "seconds": 0.8243736469994474,
+      "bytes": 1000268,
+      "pages": 30,
       "shards": 1
     },
     {
-      "seconds": 1.8200730230000772,
-      "bytes": 7217811,
-      "pages": 25,
+      "seconds": 1.0672549959999742,
+      "bytes": 7538580,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9584536009997464,
-      "bytes": 5627627,
+      "seconds": 1.1064888179989794,
+      "bytes": 5541578,
       "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.156600534000063,
-      "bytes": 13506264,
-      "pages": 39,
-      "shards": 4
+      "seconds": 0.9643621880004503,
+      "bytes": 7410728,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.817877588000556,
-      "bytes": 6663126,
-      "pages": 35,
+      "seconds": 0.8883736560001125,
+      "bytes": 6293157,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9133121940003548,
-      "bytes": 1030486,
-      "pages": 31,
+      "seconds": 0.9868076469992957,
+      "bytes": 927962,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.181683423999857,
-      "bytes": 12141179,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.2888746650005487,
+      "bytes": 18374728,
+      "pages": 40,
+      "shards": 5
     },
     {
-      "seconds": 1.1883426349995716,
-      "bytes": 1212298,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.1437757880012214,
+      "bytes": 6056576,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.1447565099988424,
-      "bytes": 5933026,
-      "pages": 35,
+      "seconds": 1.102731920998849,
+      "bytes": 2598092,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2180850740005553,
-      "bytes": 13486469,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.1366644360005012,
+      "bytes": 5837515,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.0477722719988378,
-      "bytes": 4809637,
-      "pages": 36,
+      "seconds": 0.8290303609992407,
+      "bytes": 5336251,
+      "pages": 39,
       "shards": 2
     },
     {
-      "seconds": 1.1933620469990274,
-      "bytes": 7601419,
-      "pages": 37,
+      "seconds": 1.0252480340004695,
+      "bytes": 7371694,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.0362921949999873,
-      "bytes": 5998979,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.1864866850010003,
+      "bytes": 11535625,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 0.7888650649983902,
-      "bytes": 730238,
-      "pages": 23,
-      "shards": 1
+      "seconds": 0.9566465869993408,
+      "bytes": 1137762,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.1390384229998745,
-      "bytes": 12216394,
-      "pages": 44,
+      "seconds": 1.0941519280004286,
+      "bytes": 11375746,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 0.9729804899998271,
-      "bytes": 6782943,
-      "pages": 30,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0282262380005704,
-      "bytes": 6456198,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.7203741030007222,
+      "bytes": 1052913,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.007465778999176,
-      "bytes": 7569670,
-      "pages": 36,
+      "seconds": 1.0976181589994667,
+      "bytes": 7695785,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.248841076001554,
-      "bytes": 5066284,
+      "seconds": 1.024126566000632,
+      "bytes": 5953181,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0381730779990903,
-      "bytes": 7573812,
-      "pages": 37,
+      "seconds": 1.452807390000089,
+      "bytes": 11684781,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 0.756838974000857,
-      "bytes": 1060731,
-      "pages": 33,
+      "seconds": 0.8971567329990648,
+      "bytes": 7415356,
+      "pages": 29,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9046226490008848,
+      "bytes": 1040439,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.1045795680001902,
-      "bytes": 11876139,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.043625083999359,
+      "bytes": 6212867,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.0963092559995857,
-      "bytes": 12324311,
-      "pages": 39,
-      "shards": 5
+      "seconds": 1.3019511900001817,
+      "bytes": 12069002,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 0.7185119169989775,
-      "bytes": 873551,
-      "pages": 25,
+      "seconds": 0.8234439619991463,
+      "bytes": 1074472,
+      "pages": 32,
       "shards": 1
+    },
+    {
+      "seconds": 1.0881272910010011,
+      "bytes": 13091178,
+      "pages": 33,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ad5414aa9cc5e09a287ca937f00d1a790494a4a5870ec4a0ca71e8a2bf9bfdc0
-size 721772735
+oid sha256:54b03f408ed31fec8a13f34a9c659787ebd6b16e7fa218e2cea4dd5412383f2f
+size 721776959