Qwen90M step 720000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 715000,
+  "step": 720000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "b8c254cd3e3df27a533324f081bcea4d5dc590522534c2d49abee21dbe8f10c8",
+    "model.safetensors": "982ad5bd645ce1303553abc60578335d78399b9efe72845439bb9f2e39966f3b",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "08ba2dd523ece7c3398532396b6dd7590d07e9be7249052a8b8cf4f84043c025",
+    "trainer.pt": "8c7cadbc8ada2e114351e38f2dc65770375fbd3c5571e7a3db009d13ce7e870a",
     "training_config.json": "78af77f846fb3dcba08002ed692d24a6e4b3900ba35e5592c606aca3466b41f8",
     "performance.json": "96c7d3993f53da33aeba321717c341456e5878e35efa676b6ffe17d014497e29",
-    "shard_scheduler.json": "c6173291cb813b7c0485c738e6b5e7c240ea2f582ce47fb3f2c695916a22695f",
-    "coverage.json": "e0b100ec4fa9dab8fcb0642981922fafa82749f90200b316989ed84ba5d792fc",
+    "shard_scheduler.json": "21430a5915815eb5e02a03ff691f8f7587a8b2a3d2a94f5f01d048866c5097af",
+    "coverage.json": "480e76c148c2db0ee587dc2579d82fe33abc92e0d0c035547dd68eda66cebf1c",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "1dee8333ad826e9fb3d197c33535515cdb9de560b94f624cae15e044b464f0ac"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 715000,
+  "step": 720000,
   "stats": {
-    "climbmix/blocks": 25740000,
-    "climbmix/targets": 43029519747,
-    "climbmix/input_tokens": 43032246779,
-    "ultra_qa/blocks": 9652500,
-    "ultra_qa/targets": 15747139609,
-    "ultra_qa/input_tokens": 15747170018,
-    "cortex/blocks": 9652500,
-    "cortex/targets": 367020487,
-    "cortex/input_tokens": 15343129639,
-    "ultra_style/blocks": 9652500,
-    "ultra_style/targets": 15267128059,
-    "ultra_style/input_tokens": 15267929164,
-    "openmath/blocks": 6435000,
-    "openmath/targets": 9943322469,
-    "openmath/input_tokens": 9946635258,
-    "rewrite6/blocks": 3217500,
-    "rewrite6/targets": 5274681721,
-    "rewrite6/input_tokens": 5280464589
+    "climbmix/blocks": 25920000,
+    "climbmix/targets": 43330222714,
+    "climbmix/input_tokens": 43332969001,
+    "ultra_qa/blocks": 9720000,
+    "ultra_qa/targets": 15857175972,
+    "ultra_qa/input_tokens": 15857206570,
+    "cortex/blocks": 9720000,
+    "cortex/targets": 369582035,
+    "cortex/input_tokens": 15450455935,
+    "ultra_style/blocks": 9720000,
+    "ultra_style/targets": 15374177136,
+    "ultra_style/input_tokens": 15374983563,
+    "openmath/blocks": 6480000,
+    "openmath/targets": 10012814408,
+    "openmath/input_tokens": 10016150211,
+    "rewrite6/blocks": 3240000,
+    "rewrite6/targets": 5311509903,
+    "rewrite6/input_tokens": 5317333313
   },
   "sources": {
     "climbmix": {
-      "page": 3525,
-      "row": 1183,
+      "page": 3550,
+      "row": 129,
       "start": {
-        "file": 170,
+        "file": 171,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 50,
+          "row_group": 68,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 38912,
-      "documents": 14439583
+      "offset": 0,
+      "documents": 14540929
     },
     "rewrite6": {
-      "page": 5,
-      "row": 1882,
+      "page": 19,
+      "row": 2309,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 20480
+          "row": 77824
         }
       },
       "epoch": 40,
-      "offset": 2048,
-      "documents": 8238522
+      "offset": 0,
+      "documents": 8296293
     },
     "openmath": {
-      "page": 2203,
-      "row": 534,
+      "page": 2218,
+      "row": 1902,
       "start": {
         "file": 20,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 290,
-          "row": 488
+          "row_group": 351,
+          "row": 928
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 9024022
+      "offset": 0,
+      "documents": 9086830
     },
     "ultra_qa": {
-      "page": 952,
-      "row": 2369,
+      "page": 959,
+      "row": 986,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 3902
+          "row": 32574
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3901761
+      "documents": 3929050
     },
     "ultra_style": {
-      "page": 1454,
-      "row": 163,
+      "page": 1464,
+      "row": 135,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 5,
-          "row": 49621
+          "row_group": 6,
+          "row": 6394
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 5955747
+      "documents": 5996679
     },
     "cortex": {
-      "page": 187198,
-      "row": 117,
+      "page": 188509,
+      "row": 47,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 23961461
+      "documents": 24129199
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 715000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "b8c254cd3e3df27a533324f081bcea4d5dc590522534c2d49abee21dbe8f10c8", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "08ba2dd523ece7c3398532396b6dd7590d07e9be7249052a8b8cf4f84043c025", "training_config.json": "78af77f846fb3dcba08002ed692d24a6e4b3900ba35e5592c606aca3466b41f8", "performance.json": "96c7d3993f53da33aeba321717c341456e5878e35efa676b6ffe17d014497e29", "shard_scheduler.json": "c6173291cb813b7c0485c738e6b5e7c240ea2f582ce47fb3f2c695916a22695f", "coverage.json": "e0b100ec4fa9dab8fcb0642981922fafa82749f90200b316989ed84ba5d792fc", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "1dee8333ad826e9fb3d197c33535515cdb9de560b94f624cae15e044b464f0ac", "folder": "step-000000715000"}
+{"step": 720000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "982ad5bd645ce1303553abc60578335d78399b9efe72845439bb9f2e39966f3b", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "8c7cadbc8ada2e114351e38f2dc65770375fbd3c5571e7a3db009d13ce7e870a", "training_config.json": "78af77f846fb3dcba08002ed692d24a6e4b3900ba35e5592c606aca3466b41f8", "performance.json": "96c7d3993f53da33aeba321717c341456e5878e35efa676b6ffe17d014497e29", "shard_scheduler.json": "21430a5915815eb5e02a03ff691f8f7587a8b2a3d2a94f5f01d048866c5097af", "coverage.json": "480e76c148c2db0ee587dc2579d82fe33abc92e0d0c035547dd68eda66cebf1c", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "1dee8333ad826e9fb3d197c33535515cdb9de560b94f624cae15e044b464f0ac", "folder": "step-000000720000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8c254cd3e3df27a533324f081bcea4d5dc590522534c2d49abee21dbe8f10c8
+oid sha256:982ad5bd645ce1303553abc60578335d78399b9efe72845439bb9f2e39966f3b
 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": 77176.89841787542,
-      "required_pages_per_second_upper_estimate": 0.005339276907624682
+      "required_token_positions_per_second": 76826.23510929271,
+      "required_pages_per_second_upper_estimate": 0.005315017206285955
     },
     {
       "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": 9647.112302234427,
-      "required_pages_per_second_upper_estimate": 0.004039044466332516
+      "required_token_positions_per_second": 9603.27938866159,
+      "required_pages_per_second_upper_estimate": 0.004020692540754891
     },
     {
       "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": 19294.224604468855,
-      "required_pages_per_second_upper_estimate": 0.004018478162191183
+      "required_token_positions_per_second": 19206.55877732318,
+      "required_pages_per_second_upper_estimate": 0.004000219682300069
     },
     {
       "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": 28941.33690670328,
-      "required_pages_per_second_upper_estimate": 0.001734988956694139
+      "required_token_positions_per_second": 28809.83816598477,
+      "required_pages_per_second_upper_estimate": 0.0017271058079750155
     },
     {
       "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": 28941.33690670328,
-      "required_pages_per_second_upper_estimate": 0.002250061645008552
+      "required_token_positions_per_second": 28809.83816598477,
+      "required_pages_per_second_upper_estimate": 0.002239838196319521
     },
     {
       "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": 28941.33690670328,
-      "required_pages_per_second_upper_estimate": 0.3712283950526966
+      "required_token_positions_per_second": 28809.83816598477,
+      "required_pages_per_second_upper_estimate": 0.36954167039910685
     }
   ],
   "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": 720750,
-    "producer_batch_seconds": 0.09608320560147376,
-    "producer_p95_seconds": 0.27539721799985273,
-    "gpu_update_seconds": 0.9553117773766794,
-    "producer_batches_per_second": 10.407646099441353,
-    "gpu_batches_per_second": 1.0467786786278674,
-    "producer_headroom": 9.942546893564783,
+    "producer_batch_seconds": 0.1271348553984808,
+    "producer_p95_seconds": 0.3210992399981478,
+    "gpu_update_seconds": 0.9596721731204818,
+    "producer_batches_per_second": 7.865663565398208,
+    "gpu_batches_per_second": 1.0420225031099815,
+    "producer_headroom": 7.548458446840295,
     "sustainable": true,
-    "buffer_seconds": 60.1846419747308,
+    "buffer_seconds": 60.459346906590355,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 84,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 397695.1736651185,
-    "upload_bytes_per_second": 4823252.873992993,
+    "generation_bytes_per_second": 345923.98800616147,
+    "upload_bytes_per_second": 5405264.117713781,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 85.33612639870803,
+    "estimated_full_shards_per_hour": 74.22723512781747,
     "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": 1.369231013002718,
-      "bytes": 10621268,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9135342390000005,
-      "bytes": 1211398,
-      "pages": 33,
+      "seconds": 1.016105377002532,
+      "bytes": 1143206,
+      "pages": 36,
       "shards": 1
     },
     {
-      "seconds": 1.2419597409971175,
-      "bytes": 13297063,
-      "pages": 34,
-      "shards": 4
+      "seconds": 4.951656708999508,
+      "bytes": 5644531,
+      "pages": 20,
+      "shards": 2
     },
     {
-      "seconds": 1.2846820170016144,
-      "bytes": 6246650,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.038082987000962,
+      "bytes": 8819669,
+      "pages": 48,
+      "shards": 3
     },
     {
-      "seconds": 1.021862665998924,
-      "bytes": 6426718,
-      "pages": 33,
+      "seconds": 1.3574339080005302,
+      "bytes": 10377609,
+      "pages": 27,
       "shards": 3
     },
     {
-      "seconds": 0.9455101860003197,
-      "bytes": 6297280,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2949794519990974,
+      "bytes": 7882340,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.347191354001552,
-      "bytes": 5982017,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.516641152997181,
+      "bytes": 1120266,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 3.3193765439973504,
-      "bytes": 2461242,
-      "pages": 33,
+      "seconds": 1.2580970729977707,
+      "bytes": 5952993,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.3424353250011336,
-      "bytes": 15776988,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.295129406000342,
+      "bytes": 6202154,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.047338389998913,
-      "bytes": 2628784,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.2203903480003646,
+      "bytes": 12408575,
+      "pages": 47,
+      "shards": 3
     },
     {
-      "seconds": 1.2893916930006526,
-      "bytes": 6001693,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.1006403920000594,
+      "bytes": 7578962,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.5835260300009395,
-      "bytes": 761598,
-      "pages": 23,
+      "seconds": 1.032797053001559,
+      "bytes": 1142518,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.654032960999757,
-      "bytes": 12595480,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.317680020998523,
+      "bytes": 10858764,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.5982389480013808,
-      "bytes": 6084209,
-      "pages": 37,
+      "seconds": 1.2862951650022296,
+      "bytes": 2384330,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.98295405799945,
-      "bytes": 6764576,
-      "pages": 29,
+      "seconds": 1.0045538510021288,
+      "bytes": 5798532,
+      "pages": 25,
       "shards": 2
     },
     {
-      "seconds": 1.2006459389995143,
-      "bytes": 7651087,
-      "pages": 39,
+      "seconds": 1.256682400999125,
+      "bytes": 12134890,
+      "pages": 46,
       "shards": 3
     },
     {
-      "seconds": 1.3339825349976309,
-      "bytes": 10020438,
-      "pages": 33,
+      "seconds": 1.024204646000726,
+      "bytes": 6281125,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 0.850007712997467,
-      "bytes": 1050252,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.1149510969989933,
+      "bytes": 6170756,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.2932650000002468,
-      "bytes": 7701652,
-      "pages": 38,
+      "seconds": 0.9807241070011514,
+      "bytes": 2284775,
+      "pages": 28,
       "shards": 3
     },
     {
-      "seconds": 1.022619493000093,
-      "bytes": 1029637,
+      "seconds": 1.0974279740003112,
+      "bytes": 5896433,
       "pages": 33,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0783612599989283,
-      "bytes": 5165514,
-      "pages": 4,
-      "shards": 2
-    },
-    {
-      "seconds": 1.7178185550001217,
-      "bytes": 4032890,
-      "pages": 3,
       "shards": 2
     },
     {
-      "seconds": 1.248450455001148,
-      "bytes": 5769233,
-      "pages": 1,
-      "shards": 1
+      "seconds": 1.4275493559980532,
+      "bytes": 16198066,
+      "pages": 38,
+      "shards": 4
     },
     {
-      "seconds": 1.0994338210002752,
-      "bytes": 2592435,
+      "seconds": 1.0836265130019456,
+      "bytes": 2657150,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.061899332999019,
-      "bytes": 6323038,
-      "pages": 39,
+      "seconds": 0.9289632090003579,
+      "bytes": 5852982,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.428008876999229,
-      "bytes": 7759017,
+      "seconds": 1.0356132939996314,
+      "bytes": 7513178,
       "pages": 37,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 0.8940000050024537,
-      "bytes": 393500,
-      "pages": 10,
-      "shards": 1
+      "seconds": 1.2883466929997667,
+      "bytes": 5484455,
+      "pages": 22,
+      "shards": 2
     },
     {
-      "seconds": 1.0701383749983506,
-      "bytes": 5409485,
-      "pages": 27,
+      "seconds": 1.2610901230000309,
+      "bytes": 5822650,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2911368530003529,
-      "bytes": 5955841,
-      "pages": 31,
+      "seconds": 0.9832762490004825,
+      "bytes": 2733834,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0264875259999826,
-      "bytes": 2166910,
-      "pages": 21,
-      "shards": 2
+      "seconds": 1.1357609220031009,
+      "bytes": 11893185,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.2000135250018502,
-      "bytes": 5803444,
-      "pages": 1,
+      "seconds": 0.8795082219985488,
+      "bytes": 997738,
+      "pages": 30,
       "shards": 1
     },
     {
-      "seconds": 1.2405726069991942,
-      "bytes": 6889170,
-      "pages": 43,
+      "seconds": 1.1275165429979097,
+      "bytes": 11698586,
+      "pages": 39,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0235819969966542,
+      "bytes": 6134782,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9624333549982111,
-      "bytes": 484343,
-      "pages": 14,
+      "seconds": 1.3678917270008242,
+      "bytes": 1154643,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.4119799629988847,
-      "bytes": 11493188,
-      "pages": 37,
+      "seconds": 1.2100134370048181,
+      "bytes": 12201541,
+      "pages": 35,
       "shards": 4
     },
     {
-      "seconds": 0.9430077399993024,
-      "bytes": 1135014,
+      "seconds": 1.149907068000175,
+      "bytes": 11779236,
       "pages": 33,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 1.1626808530018025,
-      "bytes": 5979744,
-      "pages": 31,
+      "seconds": 1.051513154998247,
+      "bytes": 2589054,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.2183306759980042,
-      "bytes": 7063267,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.0851055740058655,
+      "bytes": 5876732,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.0798259360017255,
-      "bytes": 4403386,
-      "pages": 18,
-      "shards": 2
+      "seconds": 1.0857899119946524,
+      "bytes": 1135012,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 1.2122206399981224,
-      "bytes": 13761400,
-      "pages": 50,
+      "seconds": 1.3140387780003948,
+      "bytes": 11211496,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 1.1497695119978744,
-      "bytes": 863102,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.1181462930035195,
+      "bytes": 5980396,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.423836862002645,
-      "bytes": 10998784,
-      "pages": 40,
+      "seconds": 1.134683780001069,
+      "bytes": 8428323,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.2024595770017186,
-      "bytes": 7444430,
+      "seconds": 1.2588409719974152,
+      "bytes": 10473930,
       "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.016105377002532,
-      "bytes": 1143206,
-      "pages": 36,
+      "seconds": 0.8643372480000835,
+      "bytes": 910531,
+      "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 4.951656708999508,
-      "bytes": 5644531,
-      "pages": 20,
+      "seconds": 1.064887002001342,
+      "bytes": 7934391,
+      "pages": 43,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0906057760003023,
+      "bytes": 6005491,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.038082987000962,
-      "bytes": 8819669,
-      "pages": 48,
-      "shards": 3
+      "seconds": 0.9286484479962382,
+      "bytes": 714826,
+      "pages": 21,
+      "shards": 1
     },
     {
-      "seconds": 1.3574339080005302,
-      "bytes": 10377609,
-      "pages": 27,
-      "shards": 3
+      "seconds": 1.5466609349969076,
+      "bytes": 4766956,
+      "pages": 1,
+      "shards": 1
     },
     {
-      "seconds": 1.2949794519990974,
-      "bytes": 7882340,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.2835849650000455,
+      "bytes": 14052046,
+      "pages": 50,
+      "shards": 4
     },
     {
-      "seconds": 1.516641152997181,
-      "bytes": 1120266,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.258647842994833,
+      "bytes": 5993179,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.2580970729977707,
-      "bytes": 5952993,
-      "pages": 28,
+      "seconds": 1.1118833660002565,
+      "bytes": 2568385,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.295129406000342,
-      "bytes": 6202154,
-      "pages": 29,
+      "seconds": 0.9927107309995336,
+      "bytes": 6005616,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.2203903480003646,
-      "bytes": 12408575,
-      "pages": 47,
+      "seconds": 1.3781777679978404,
+      "bytes": 5044232,
+      "pages": 36,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3652096680016257,
+      "bytes": 7517328,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.1006403920000594,
-      "bytes": 7578962,
-      "pages": 31,
+      "seconds": 1.3108274569967762,
+      "bytes": 11956950,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.032797053001559,
-      "bytes": 1142518,
-      "pages": 33,
+      "seconds": 0.923527514001762,
+      "bytes": 789312,
+      "pages": 22,
       "shards": 1
     },
     {
-      "seconds": 1.317680020998523,
-      "bytes": 10858764,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.4157909589994233,
+      "bytes": 12468598,
+      "pages": 49,
+      "shards": 4
     },
     {
-      "seconds": 1.2862951650022296,
-      "bytes": 2384330,
-      "pages": 29,
-      "shards": 2
+      "seconds": 0.873544275003951,
+      "bytes": 984851,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.0045538510021288,
-      "bytes": 5798532,
-      "pages": 25,
+      "seconds": 1.161854639001831,
+      "bytes": 7666603,
+      "pages": 40,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9383765300008235,
+      "bytes": 5995522,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.256682400999125,
-      "bytes": 12134890,
-      "pages": 46,
+      "seconds": 1.2260548230042332,
+      "bytes": 10646786,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.024204646000726,
-      "bytes": 6281125,
-      "pages": 32,
+      "seconds": 1.1512752330017975,
+      "bytes": 7594320,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.1149510969989933,
-      "bytes": 6170756,
-      "pages": 37,
+      "seconds": 1.0619875439952011,
+      "bytes": 5471838,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9807241070011514,
-      "bytes": 2284775,
-      "pages": 28,
+      "seconds": 1.1333524080037023,
+      "bytes": 7101020,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.0974279740003112,
-      "bytes": 5896433,
-      "pages": 33,
+      "seconds": 1.235225882002851,
+      "bytes": 5723234,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.4275493559980532,
-      "bytes": 16198066,
-      "pages": 38,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0836265130019456,
-      "bytes": 2657150,
-      "pages": 38,
+      "seconds": 0.9995385679940227,
+      "bytes": 6053776,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9289632090003579,
-      "bytes": 5852982,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.3147696169980918,
+      "bytes": 8376347,
+      "pages": 36,
+      "shards": 3
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:08ba2dd523ece7c3398532396b6dd7590d07e9be7249052a8b8cf4f84043c025
-size 721776063
+oid sha256:8c7cadbc8ada2e114351e38f2dc65770375fbd3c5571e7a3db009d13ce7e870a
+size 721767167