Qwen90M step 255000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 250000,
+  "step": 255000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "b39daf23cbfc7a4db45dc715dda0257d7a0aceacebd6722e70e52b392a3d914f",
+    "model.safetensors": "fd97f7dff3c950a4c906fc0d4a7fd3b2954ec1ce69c8fc7fe9fbf0fb58cc2fc1",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "e91afb32a91ba214e7473f46d6f99a0a931e405920d457559aea60d50123c7e4",
+    "trainer.pt": "a2ebbe485478b945d252786a3ac085f24afc9d7d83bc13cc75393c2a988fb4dd",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "ac15a69eb79c2ac9ef8406edc12d9ae8abafb561b0f74b54b23d5920ee4fe3f7",
-    "shard_scheduler.json": "51e06602336cf2ff9c7fbf2702cb9edfe3a4879775d6e82cb900f0895494a34a",
-    "coverage.json": "fa6f7ebd74b32c3a670cb83966d87b06345ed8c181b0803237a454541a8aa878",
+    "shard_scheduler.json": "040240fbf545ea1d490b42b841ab997a911ae28bae5ed7a72035f5d7db770fd7",
+    "coverage.json": "836ac61bfc32d5771b0a24e2739ea3cc1013fa5a041bce0fbb429b3c8bf253f8",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 250000,
+  "step": 255000,
   "stats": {
-    "climbmix/blocks": 9000000,
-    "climbmix/targets": 15044821319,
-    "climbmix/input_tokens": 15045774685,
-    "ultra_qa/blocks": 3375000,
-    "ultra_qa/targets": 5505900381,
-    "ultra_qa/input_tokens": 5505910963,
-    "cortex/blocks": 3375000,
-    "cortex/targets": 128400391,
-    "cortex/input_tokens": 5364742698,
-    "ultra_style/blocks": 3375000,
-    "ultra_style/targets": 5338964418,
-    "ultra_style/input_tokens": 5339244172,
-    "openmath/blocks": 2250000,
-    "openmath/targets": 3476637803,
-    "openmath/input_tokens": 3477796696,
-    "rewrite6/blocks": 1125000,
-    "rewrite6/targets": 1844270111,
-    "rewrite6/input_tokens": 1846292063
+    "climbmix/blocks": 9180000,
+    "climbmix/targets": 15346078422,
+    "climbmix/input_tokens": 15347051076,
+    "ultra_qa/blocks": 3442500,
+    "ultra_qa/targets": 5615917971,
+    "ultra_qa/input_tokens": 5615928771,
+    "cortex/blocks": 3442500,
+    "cortex/targets": 130960799,
+    "cortex/input_tokens": 5472030778,
+    "ultra_style/blocks": 3442500,
+    "ultra_style/targets": 5445625285,
+    "ultra_style/input_tokens": 5445910748,
+    "openmath/blocks": 2295000,
+    "openmath/targets": 3546217432,
+    "openmath/input_tokens": 3547399641,
+    "rewrite6/blocks": 1147500,
+    "rewrite6/targets": 1881142352,
+    "rewrite6/input_tokens": 1883204831
   },
   "sources": {
     "climbmix": {
-      "page": 1232,
-      "row": 2049,
+      "page": 1257,
+      "row": 595,
       "start": {
-        "file": 59,
+        "file": 60,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 53,
+          "row_group": 70,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 5048321
+      "documents": 5149267
     },
     "rewrite6": {
-      "page": 1,
-      "row": 847,
+      "page": 15,
+      "row": 1160,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 4096
+          "row": 61440
         }
       },
       "epoch": 14,
       "offset": 0,
-      "documents": 2880599
+      "documents": 2938256
     },
     "openmath": {
-      "page": 770,
-      "row": 2180,
+      "page": 785,
+      "row": 3990,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 97,
-          "row": 370
+          "row_group": 158,
+          "row": 810
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 3156100
+      "documents": 3219350
     },
     "ultra_qa": {
-      "page": 333,
-      "row": 364,
+      "page": 339,
+      "row": 3105,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 1742
+          "row": 26318
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1364332
+      "documents": 1391649
     },
     "ultra_style": {
-      "page": 508,
-      "row": 1215,
+      "page": 518,
+      "row": 2074,
       "start": {
         "file": 3,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 0,
-          "row": 25940
+          "row": 66900
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 2081983
+      "documents": 2123802
     },
     "cortex": {
-      "page": 65453,
-      "row": 34,
+      "page": 66763,
+      "row": 12,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 8378018
+      "documents": 8545676
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 250000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "b39daf23cbfc7a4db45dc715dda0257d7a0aceacebd6722e70e52b392a3d914f", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "e91afb32a91ba214e7473f46d6f99a0a931e405920d457559aea60d50123c7e4", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "ac15a69eb79c2ac9ef8406edc12d9ae8abafb561b0f74b54b23d5920ee4fe3f7", "shard_scheduler.json": "51e06602336cf2ff9c7fbf2702cb9edfe3a4879775d6e82cb900f0895494a34a", "coverage.json": "fa6f7ebd74b32c3a670cb83966d87b06345ed8c181b0803237a454541a8aa878", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000250000"}
+{"step": 255000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "fd97f7dff3c950a4c906fc0d4a7fd3b2954ec1ce69c8fc7fe9fbf0fb58cc2fc1", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "a2ebbe485478b945d252786a3ac085f24afc9d7d83bc13cc75393c2a988fb4dd", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "ac15a69eb79c2ac9ef8406edc12d9ae8abafb561b0f74b54b23d5920ee4fe3f7", "shard_scheduler.json": "040240fbf545ea1d490b42b841ab997a911ae28bae5ed7a72035f5d7db770fd7", "coverage.json": "836ac61bfc32d5771b0a24e2739ea3cc1013fa5a041bce0fbb429b3c8bf253f8", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000255000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b39daf23cbfc7a4db45dc715dda0257d7a0aceacebd6722e70e52b392a3d914f
+oid sha256:fd97f7dff3c950a4c906fc0d4a7fd3b2954ec1ce69c8fc7fe9fbf0fb58cc2fc1
 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": 75570.71051583829,
-      "required_pages_per_second_upper_estimate": 0.006550277655298134
+      "required_token_positions_per_second": 75186.44264578017,
+      "required_pages_per_second_upper_estimate": 0.006516970290239508
     },
     {
       "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": 9446.338814479786,
-      "required_pages_per_second_upper_estimate": 0.003768285788447338
+      "required_token_positions_per_second": 9398.305330722522,
+      "required_pages_per_second_upper_estimate": 0.003749124513611984
     },
     {
       "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": 18892.677628959573,
-      "required_pages_per_second_upper_estimate": 0.004370488469689805
+      "required_token_positions_per_second": 18796.610661445044,
+      "required_pages_per_second_upper_estimate": 0.004348265067476201
     },
     {
       "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": 28339.01644343936,
-      "required_pages_per_second_upper_estimate": 0.0017256974180770313
+      "required_token_positions_per_second": 28194.915992167564,
+      "required_pages_per_second_upper_estimate": 0.0017169224566312173
     },
     {
       "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": 28339.01644343936,
-      "required_pages_per_second_upper_estimate": 0.0021816159332343353
+      "required_token_positions_per_second": 28194.915992167564,
+      "required_pages_per_second_upper_estimate": 0.002170522681599853
     },
     {
       "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": 28339.01644343936,
-      "required_pages_per_second_upper_estimate": 0.3518757396406541
+      "required_token_positions_per_second": 28194.915992167564,
+      "required_pages_per_second_upper_estimate": 0.3500864943090451
     }
   ],
   "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": 690670,
-    "producer_batch_seconds": 0.129124772499992,
-    "producer_p95_seconds": 0.24966627399953722,
-    "gpu_update_seconds": 0.9756160752855156,
-    "producer_batches_per_second": 7.744447332908656,
-    "gpu_batches_per_second": 1.0249933609461575,
-    "producer_headroom": 7.555607312187723,
+    "producer_batch_seconds": 0.08646610254689335,
+    "producer_p95_seconds": 0.24833947200022521,
+    "gpu_update_seconds": 0.9806023187896891,
+    "producer_batches_per_second": 11.565225799991017,
+    "gpu_batches_per_second": 1.0197813943926346,
+    "producer_headroom": 11.340887236797528,
     "sustainable": true,
-    "buffer_seconds": 60.48819666770197,
+    "buffer_seconds": 60.79734376496072,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 78,
+    "upload_pages_per_commit": 84,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 382211.29352878686,
-    "upload_bytes_per_second": 4854486.8576622605,
+    "generation_bytes_per_second": 312898.0981319677,
+    "upload_bytes_per_second": 4945666.70118425,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 82.01364616773323,
+    "estimated_full_shards_per_hour": 67.14064796418451,
     "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": 0.7338976749997528,
-      "bytes": 1218142,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1807161469996572,
-      "bytes": 7525394,
-      "pages": 28,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1362227980007447,
-      "bytes": 11027017,
-      "pages": 35,
+      "seconds": 4.401627224999174,
+      "bytes": 7622267,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 0.8288988310005152,
-      "bytes": 5801101,
-      "pages": 35,
+      "seconds": 0.9423913660011749,
+      "bytes": 5379592,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.2185628970000835,
-      "bytes": 7809668,
+      "seconds": 1.2426582910011348,
+      "bytes": 7714336,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.2807725910006411,
-      "bytes": 1082933,
-      "pages": 33,
+      "seconds": 1.0709264350007288,
+      "bytes": 727110,
+      "pages": 23,
       "shards": 1
     },
     {
-      "seconds": 1.2944532310002614,
-      "bytes": 11460253,
+      "seconds": 1.4975259919992823,
+      "bytes": 12239566,
+      "pages": 44,
+      "shards": 3
+    },
+    {
+      "seconds": 1.4755159030009963,
+      "bytes": 11647558,
       "pages": 33,
       "shards": 4
     },
     {
-      "seconds": 1.090632098000242,
-      "bytes": 1122627,
+      "seconds": 0.8895111490001,
+      "bytes": 1197830,
       "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.176693084999897,
-      "bytes": 12183576,
-      "pages": 36,
+      "seconds": 1.169628391000515,
+      "bytes": 7496810,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0236794259999442,
-      "bytes": 7487764,
-      "pages": 29,
+      "seconds": 1.5263510319982743,
+      "bytes": 5728692,
+      "pages": 27,
       "shards": 3
     },
     {
-      "seconds": 1.1006291130006502,
-      "bytes": 6137509,
-      "pages": 39,
+      "seconds": 1.2046007990011276,
+      "bytes": 5154137,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.0797216399996614,
-      "bytes": 5987959,
-      "pages": 27,
-      "shards": 2
+      "seconds": 2.8337960279986874,
+      "bytes": 11685666,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.322158891000072,
-      "bytes": 2479484,
+      "seconds": 1.282873722999284,
+      "bytes": 2666361,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.3229940370001714,
-      "bytes": 6465970,
-      "pages": 36,
+      "seconds": 1.1408444070002588,
+      "bytes": 5962047,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.3629516780001723,
-      "bytes": 11728657,
-      "pages": 30,
+      "seconds": 1.150664095999673,
+      "bytes": 11982825,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 1.096408940999936,
-      "bytes": 6091544,
+      "seconds": 0.9085387359991728,
+      "bytes": 1084848,
+      "pages": 33,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2088771979997546,
+      "bytes": 6082441,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.042788374000338,
-      "bytes": 6163555,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.0392706679995172,
+      "bytes": 8238622,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.04696464799963,
-      "bytes": 2343472,
-      "pages": 28,
+      "seconds": 1.0565556260007725,
+      "bytes": 5902213,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.3791698659997564,
-      "bytes": 10729274,
-      "pages": 36,
+      "seconds": 1.6428124459998799,
+      "bytes": 11197872,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 1.2665546050002376,
-      "bytes": 6293373,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4158054800000173,
-      "bytes": 8082619,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.1795239030016091,
+      "bytes": 829951,
+      "pages": 25,
+      "shards": 1
     },
     {
-      "seconds": 1.100743914999839,
-      "bytes": 6375983,
+      "seconds": 1.183389565001562,
+      "bytes": 6316614,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1518813020011294,
-      "bytes": 6877765,
-      "pages": 35,
+      "seconds": 1.805858101000922,
+      "bytes": 7507792,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.007327716999498,
-      "bytes": 296016,
-      "pages": 9,
-      "shards": 1
+      "seconds": 2.5084295430006023,
+      "bytes": 11998594,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.1130705590003345,
-      "bytes": 6042311,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.094205030000012,
+      "bytes": 275280,
+      "pages": 8,
+      "shards": 1
     },
     {
-      "seconds": 0.9439528090006206,
-      "bytes": 7900957,
+      "seconds": 1.677992199000073,
+      "bytes": 7646685,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.7852815879996342,
-      "bytes": 838334,
-      "pages": 26,
+      "seconds": 0.9418184709993511,
+      "bytes": 1092329,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 4.030324741001095,
-      "bytes": 15386927,
-      "pages": 28,
-      "shards": 4
-    },
-    {
-      "seconds": 0.845626510999864,
-      "bytes": 2613623,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.2206816289999551,
+      "bytes": 9827767,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.1139741530005267,
-      "bytes": 5909960,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.1708706409990555,
+      "bytes": 7164779,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.2427591689993278,
-      "bytes": 910923,
-      "pages": 26,
+      "seconds": 1.2709880140009773,
+      "bytes": 1216081,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.2945012599993788,
-      "bytes": 12063808,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.5031630230005248,
+      "bytes": 11828419,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 0.8872203939990868,
-      "bytes": 6084404,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.1863015779999841,
+      "bytes": 6891470,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.511142709001433,
-      "bytes": 6655300,
-      "pages": 29,
+      "seconds": 1.3068818119991192,
+      "bytes": 5782255,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0988184660000115,
-      "bytes": 7486365,
+      "seconds": 1.4370965530015383,
+      "bytes": 7562097,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.2685245129996474,
-      "bytes": 10318034,
-      "pages": 29,
+      "seconds": 0.8747751649989368,
+      "bytes": 1118624,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1147228019999602,
+      "bytes": 10226673,
+      "pages": 28,
       "shards": 3
     },
     {
-      "seconds": 1.292125334000957,
-      "bytes": 1081552,
-      "pages": 33,
+      "seconds": 1.303761271001349,
+      "bytes": 13900408,
+      "pages": 43,
+      "shards": 4
+    },
+    {
+      "seconds": 0.9504027439998026,
+      "bytes": 830160,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 1.197562700999697,
-      "bytes": 7759288,
-      "pages": 35,
+      "seconds": 1.2181768209993606,
+      "bytes": 7539545,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.8769548000000214,
-      "bytes": 1135557,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.3715665409999929,
+      "bytes": 4880519,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 3.838755348999257,
-      "bytes": 11603941,
-      "pages": 24,
+      "seconds": 1.2786054230000445,
+      "bytes": 6003136,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.351524421999784,
+      "bytes": 7384349,
+      "pages": 30,
       "shards": 4
     },
     {
-      "seconds": 1.57969898599913,
-      "bytes": 12094782,
-      "pages": 45,
-      "shards": 3
+      "seconds": 1.2958617730000697,
+      "bytes": 7155083,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 0.8754016399998363,
-      "bytes": 1113722,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.0062387039997702,
+      "bytes": 10437075,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 4.401627224999174,
-      "bytes": 7622267,
-      "pages": 32,
+      "seconds": 1.1012680910007475,
+      "bytes": 7471820,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9423913660011749,
-      "bytes": 5379592,
+      "seconds": 0.9738275679992512,
+      "bytes": 1102759,
       "pages": 34,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 1.2426582910011348,
-      "bytes": 7714336,
-      "pages": 37,
+      "seconds": 1.0251637159999518,
+      "bytes": 7585581,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.0709264350007288,
-      "bytes": 727110,
-      "pages": 23,
-      "shards": 1
+      "seconds": 1.1754312260000006,
+      "bytes": 5287839,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.4975259919992823,
-      "bytes": 12239566,
-      "pages": 44,
+      "seconds": 1.096052540000528,
+      "bytes": 12098645,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.4755159030009963,
-      "bytes": 11647558,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.8189289659967471,
+      "bytes": 7675879,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 0.8895111490001,
-      "bytes": 1197830,
-      "pages": 33,
+      "seconds": 1.221604631999071,
+      "bytes": 830309,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.169628391000515,
-      "bytes": 7496810,
+      "seconds": 1.7278612150003028,
+      "bytes": 10862160,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.5263510319982743,
-      "bytes": 5728692,
-      "pages": 27,
-      "shards": 3
+      "seconds": 0.8819487159998971,
+      "bytes": 2668679,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.2046007990011276,
-      "bytes": 5154137,
-      "pages": 31,
+      "seconds": 1.1101788160012802,
+      "bytes": 5690031,
+      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 2.8337960279986874,
-      "bytes": 11685666,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.5474547390003863,
+      "bytes": 13326359,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.282873722999284,
-      "bytes": 2666361,
-      "pages": 35,
+      "seconds": 1.2802494179995847,
+      "bytes": 4852642,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1408444070002588,
-      "bytes": 5962047,
+      "seconds": 1.42962039700069,
+      "bytes": 6308465,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.150664095999673,
-      "bytes": 11982825,
-      "pages": 34,
-      "shards": 4
-    },
-    {
-      "seconds": 0.9085387359991728,
-      "bytes": 1084848,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.0526814939985343,
+      "bytes": 2354967,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.2088771979997546,
-      "bytes": 6082441,
+      "seconds": 1.0578431400026602,
+      "bytes": 6390274,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.0392706679995172,
-      "bytes": 8238622,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0565556260007725,
-      "bytes": 5902213,
-      "pages": 31,
+      "seconds": 1.1717372180028178,
+      "bytes": 4762409,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.6428124459998799,
-      "bytes": 11197872,
-      "pages": 39,
+      "seconds": 1.406141793999268,
+      "bytes": 13345421,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 1.1795239030016091,
-      "bytes": 829951,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.0539056280031218,
+      "bytes": 6000362,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.183389565001562,
-      "bytes": 6316614,
-      "pages": 35,
+      "seconds": 1.0857812779977394,
+      "bytes": 2115915,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.805858101000922,
-      "bytes": 7507792,
-      "pages": 34,
+      "seconds": 1.3246665000006033,
+      "bytes": 11374746,
+      "pages": 45,
       "shards": 3
+    },
+    {
+      "seconds": 1.0599599050001416,
+      "bytes": 875478,
+      "pages": 25,
+      "shards": 1
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e91afb32a91ba214e7473f46d6f99a0a931e405920d457559aea60d50123c7e4
-size 721772479
+oid sha256:a2ebbe485478b945d252786a3ac085f24afc9d7d83bc13cc75393c2a988fb4dd
+size 721771711