Qwen90M step 505000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 500000,
+  "step": 505000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "b32b5405757b9ffe78293bedfdb956dae40e8488a1c520be2cde0c5aed95c84e",
+    "model.safetensors": "e9099c8a8b03dc9572c97e934128304524a8a458a8efa4493e025952c068ceab",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "442d5a63837b6ac40b1f2659fb1a8dca8d13b6c8efafd35b7dd88a3f95501c4e",
+    "trainer.pt": "b8e924ba03c8c90f3a051947fd0eb5994e5b786f87c64a7022243d5bfe8f0021",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "234744063e094dfd7b525da4e181ac493fac400c830f871c7e8c59127ed10dc9",
-    "shard_scheduler.json": "28346349a8e1d8ef3ce4b1c69a6243a3f826cee646b53fd277baf701c2210e6e",
-    "coverage.json": "19aa4f8d54332a074a778e21560e24dcd35db2f280ad9c717dbc173725ed6762",
+    "shard_scheduler.json": "58df431fac6f1fc091d15d01fbea31a13160b04a17771a23da2d9dd054165338",
+    "coverage.json": "a92539a3090e14fb0565a5f8b4bf84185ba11d618ff984ee9c273425d8c514dd",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 500000,
+  "step": 505000,
   "stats": {
-    "climbmix/blocks": 18000000,
-    "climbmix/targets": 30087292818,
-    "climbmix/input_tokens": 30089199837,
-    "ultra_qa/blocks": 6750000,
-    "ultra_qa/targets": 11011602216,
-    "ultra_qa/input_tokens": 11011623449,
-    "cortex/blocks": 6750000,
-    "cortex/targets": 256714249,
-    "cortex/input_tokens": 10729951557,
-    "ultra_style/blocks": 6750000,
-    "ultra_style/targets": 10676414105,
-    "ultra_style/input_tokens": 10676973780,
-    "openmath/blocks": 4500000,
-    "openmath/targets": 6953343610,
-    "openmath/input_tokens": 6955660259,
-    "rewrite6/blocks": 2250000,
-    "rewrite6/targets": 3688563295,
-    "rewrite6/input_tokens": 3692607180
+    "climbmix/blocks": 18180000,
+    "climbmix/targets": 30388475981,
+    "climbmix/input_tokens": 30390402108,
+    "ultra_qa/blocks": 6817500,
+    "ultra_qa/targets": 11121758760,
+    "ultra_qa/input_tokens": 11121780202,
+    "cortex/blocks": 6817500,
+    "cortex/targets": 259282200,
+    "cortex/input_tokens": 10837250165,
+    "ultra_style/blocks": 6817500,
+    "ultra_style/targets": 10783164447,
+    "ultra_style/input_tokens": 10783729590,
+    "openmath/blocks": 4545000,
+    "openmath/targets": 7022833549,
+    "openmath/input_tokens": 7025173354,
+    "rewrite6/blocks": 2272500,
+    "rewrite6/targets": 3725426609,
+    "rewrite6/input_tokens": 3729511162
   },
   "sources": {
     "climbmix": {
-      "page": 2465,
-      "row": 3704,
+      "page": 2490,
+      "row": 2231,
       "start": {
-        "file": 119,
+        "file": 120,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 22,
+          "row_group": 39,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 10100344
+      "offset": 10240,
+      "documents": 10201271
     },
     "rewrite6": {
-      "page": 2,
-      "row": 1645,
+      "page": 16,
+      "row": 2127,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 8192
+          "row": 65536
         }
       },
       "epoch": 28,
       "offset": 0,
-      "documents": 5761149
+      "documents": 5818975
     },
     "openmath": {
-      "page": 1540,
-      "row": 2675,
+      "page": 1556,
+      "row": 257,
       "start": {
         "file": 14,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 194,
-          "row": 740
+          "row_group": 260,
+          "row": 276
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 6310515
+      "documents": 6373633
     },
     "ultra_qa": {
-      "page": 666,
-      "row": 700,
+      "page": 672,
+      "row": 3323,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 2,
-          "row": 775
+          "row": 25351
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 2728636
+      "offset": 0,
+      "documents": 2755835
     },
     "ultra_style": {
-      "page": 1016,
-      "row": 2694,
+      "page": 1026,
+      "row": 2979,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 0,
-          "row": 51884
+          "row_group": 1,
+          "row": 7627
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 4164230
+      "offset": 0,
+      "documents": 4205475
     },
     "cortex": {
-      "page": 130914,
-      "row": 26,
+      "page": 132224,
+      "row": 58,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 16757018
+      "documents": 16924730
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 500000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "b32b5405757b9ffe78293bedfdb956dae40e8488a1c520be2cde0c5aed95c84e", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "442d5a63837b6ac40b1f2659fb1a8dca8d13b6c8efafd35b7dd88a3f95501c4e", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "234744063e094dfd7b525da4e181ac493fac400c830f871c7e8c59127ed10dc9", "shard_scheduler.json": "28346349a8e1d8ef3ce4b1c69a6243a3f826cee646b53fd277baf701c2210e6e", "coverage.json": "19aa4f8d54332a074a778e21560e24dcd35db2f280ad9c717dbc173725ed6762", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000500000"}
+{"step": 505000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e9099c8a8b03dc9572c97e934128304524a8a458a8efa4493e025952c068ceab", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "b8e924ba03c8c90f3a051947fd0eb5994e5b786f87c64a7022243d5bfe8f0021", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "234744063e094dfd7b525da4e181ac493fac400c830f871c7e8c59127ed10dc9", "shard_scheduler.json": "58df431fac6f1fc091d15d01fbea31a13160b04a17771a23da2d9dd054165338", "coverage.json": "a92539a3090e14fb0565a5f8b4bf84185ba11d618ff984ee9c273425d8c514dd", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000505000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b32b5405757b9ffe78293bedfdb956dae40e8488a1c520be2cde0c5aed95c84e
+oid sha256:e9099c8a8b03dc9572c97e934128304524a8a458a8efa4493e025952c068ceab
 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": 76297.59507963236,
-      "required_pages_per_second_upper_estimate": 0.005278444662420188
+      "required_token_positions_per_second": 76710.58553286047,
+      "required_pages_per_second_upper_estimate": 0.005307016300244384
     },
     {
       "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": 9537.199384954045,
-      "required_pages_per_second_upper_estimate": 0.0039930262231099335
+      "required_token_positions_per_second": 9588.823191607558,
+      "required_pages_per_second_upper_estimate": 0.004014640032928091
     },
     {
       "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": 19074.39876990809,
-      "required_pages_per_second_upper_estimate": 0.003972694238049278
+      "required_token_positions_per_second": 19177.646383215117,
+      "required_pages_per_second_upper_estimate": 0.003994197993078467
     },
     {
       "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": 28611.598154862135,
-      "required_pages_per_second_upper_estimate": 0.0017152216220031838
+      "required_token_positions_per_second": 28766.469574822677,
+      "required_pages_per_second_upper_estimate": 0.0017245059271548526
     },
     {
       "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": 28611.598154862135,
-      "required_pages_per_second_upper_estimate": 0.0022244259074203927
+      "required_token_positions_per_second": 28766.469574822677,
+      "required_pages_per_second_upper_estimate": 0.0022364664791149408
     },
     {
       "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": 28611.598154862135,
-      "required_pages_per_second_upper_estimate": 0.36699886038996593
+      "required_token_positions_per_second": 28766.469574822677,
+      "required_pages_per_second_upper_estimate": 0.36898538467724473
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 704566,
-    "producer_batch_seconds": 0.13252937544521615,
-    "producer_p95_seconds": 0.25005536200114875,
-    "gpu_update_seconds": 0.9663214144960867,
-    "producer_batches_per_second": 7.545496963526938,
-    "gpu_batches_per_second": 1.0348523638187983,
-    "producer_headroom": 7.291375298871277,
+    "producer_batch_seconds": 0.0882111206484808,
+    "producer_p95_seconds": 0.2659228590018756,
+    "gpu_update_seconds": 0.9611189836168983,
+    "producer_batches_per_second": 11.336439132033885,
+    "gpu_batches_per_second": 1.0404539053393618,
+    "producer_headroom": 10.89566685641524,
     "sustainable": true,
-    "buffer_seconds": 60.87824911325346,
+    "buffer_seconds": 60.55049596786459,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 84,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 321426.0121983039,
-    "upload_bytes_per_second": 5933281.998072939,
+    "generation_bytes_per_second": 393537.5578407053,
+    "upload_bytes_per_second": 5534601.836951126,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 68.970539803141,
+    "estimated_full_shards_per_hour": 84.44399882713194,
     "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": 1.1922360469998239,
-      "bytes": 10997251,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0726386229998752,
-      "bytes": 2419938,
-      "pages": 28,
+      "seconds": 1.2417163159989286,
+      "bytes": 6099161,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.240178467000078,
-      "bytes": 6103593,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.1390252959990903,
+      "bytes": 7728259,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.8772162379991642,
-      "bytes": 2610307,
-      "pages": 37,
+      "seconds": 1.331549176999033,
+      "bytes": 5708517,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.183507477000603,
-      "bytes": 15672429,
-      "pages": 34,
-      "shards": 4
-    },
-    {
-      "seconds": 1.1264393319997907,
-      "bytes": 6331952,
+      "seconds": 1.0187081299991405,
+      "bytes": 6222643,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.856032101000892,
-      "bytes": 2311911,
+      "seconds": 1.2927204380011972,
+      "bytes": 2373334,
       "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.5313736190000782,
-      "bytes": 6009990,
-      "pages": 37,
+      "seconds": 1.0165509960006602,
+      "bytes": 11776139,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.0109308320006676,
-      "bytes": 5084502,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1994448559999,
+      "bytes": 10381245,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.1290311129996553,
-      "bytes": 7603249,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.1291984180006693,
+      "bytes": 2752026,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.4725222399993072,
-      "bytes": 12055069,
+      "seconds": 0.8884318259988504,
+      "bytes": 6185512,
       "pages": 36,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.2178526319985394,
-      "bytes": 7463726,
-      "pages": 32,
+      "seconds": 1.246648587000891,
+      "bytes": 6489173,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.3964880730000004,
-      "bytes": 6495644,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2645150809985353,
-      "bytes": 6000960,
+      "seconds": 1.0390239320004184,
+      "bytes": 6087872,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9461197429991444,
-      "bytes": 2470605,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.244901927000683,
+      "bytes": 11604745,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.2941537009992317,
-      "bytes": 6225550,
+      "seconds": 1.1130494770004589,
+      "bytes": 2612508,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2335469089994149,
-      "bytes": 11703737,
-      "pages": 35,
+      "seconds": 0.9836850789997698,
+      "bytes": 10345276,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1459508749994711,
-      "bytes": 7841640,
-      "pages": 34,
+      "seconds": 1.1697606620000442,
+      "bytes": 7536660,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0024716899988562,
-      "bytes": 5917383,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1789036200007104,
-      "bytes": 2313669,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.5398991070014745,
-      "bytes": 10300581,
-      "pages": 40,
-      "shards": 3
+      "seconds": 0.8494176380008867,
+      "bytes": 947075,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 0.9979469910012995,
-      "bytes": 6506963,
+      "seconds": 1.246135413999582,
+      "bytes": 6441176,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1795985310000106,
-      "bytes": 8204299,
-      "pages": 31,
+      "seconds": 0.9687062859993603,
+      "bytes": 7942604,
+      "pages": 25,
       "shards": 3
     },
     {
-      "seconds": 0.9598508790004416,
-      "bytes": 6041026,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9930732779994287,
-      "bytes": 6062259,
-      "pages": 32,
+      "seconds": 1.493546093999612,
+      "bytes": 10942153,
+      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 1.0064647640010662,
-      "bytes": 5698061,
+      "seconds": 1.185240850001719,
+      "bytes": 5667842,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1021013760000642,
-      "bytes": 5987765,
+      "seconds": 0.9565514620007889,
+      "bytes": 2665835,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0322518390003097,
-      "bytes": 2308031,
-      "pages": 30,
+      "seconds": 1.161401912002475,
+      "bytes": 5953132,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.3352466599990294,
-      "bytes": 15653746,
-      "pages": 42,
+      "seconds": 1.9063561590010067,
+      "bytes": 11430108,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 1.1028008130015223,
-      "bytes": 1043427,
-      "pages": 31,
-      "shards": 1
-    },
-    {
-      "seconds": 0.9759294559989939,
-      "bytes": 7445734,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0883692229999724,
-      "bytes": 10848676,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.1905037959986657,
+      "bytes": 6976659,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.0698365700009163,
-      "bytes": 2697855,
-      "pages": 38,
+      "seconds": 1.609685786999762,
+      "bytes": 5649551,
+      "pages": 20,
       "shards": 2
     },
     {
-      "seconds": 0.9394886710015271,
-      "bytes": 6144198,
-      "pages": 36,
+      "seconds": 0.9700031609972939,
+      "bytes": 2540670,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0910024349996092,
-      "bytes": 11590114,
-      "pages": 31,
+      "seconds": 1.1479466919990955,
+      "bytes": 10073767,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.1193175999997038,
-      "bytes": 7119189,
+      "seconds": 0.986695225998119,
+      "bytes": 1193548,
       "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.203395697997621,
+      "bytes": 7988495,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.3181563210000604,
-      "bytes": 6116642,
-      "pages": 36,
+      "seconds": 0.9280091689979599,
+      "bytes": 6513685,
+      "pages": 23,
       "shards": 2
     },
     {
-      "seconds": 0.95897385599892,
-      "bytes": 2279278,
-      "pages": 29,
-      "shards": 3
+      "seconds": 4.352831413998501,
+      "bytes": 12282177,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.0730731740004558,
-      "bytes": 6315193,
+      "seconds": 1.176046257998678,
+      "bytes": 6195296,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.276775304999319,
-      "bytes": 10081435,
+      "seconds": 0.9275896610015479,
+      "bytes": 1143867,
+      "pages": 35,
+      "shards": 1
+    },
+    {
+      "seconds": 3.292675117001636,
+      "bytes": 7453711,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.383926528000302,
+      "bytes": 10496353,
       "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.1636577470017073,
-      "bytes": 8230638,
+      "seconds": 1.054139716001373,
+      "bytes": 8196073,
       "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.2417163159989286,
-      "bytes": 6099161,
+      "seconds": 1.1359631509985775,
+      "bytes": 6053377,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1390252959990903,
-      "bytes": 7728259,
-      "pages": 38,
-      "shards": 3
+      "seconds": 0.9480016679990513,
+      "bytes": 5853547,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.331549176999033,
-      "bytes": 5708517,
-      "pages": 35,
-      "shards": 2
+      "seconds": 0.9646089939997182,
+      "bytes": 7151890,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.0187081299991405,
-      "bytes": 6222643,
+      "seconds": 1.181685737999942,
+      "bytes": 6229682,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2927204380011972,
-      "bytes": 2373334,
+      "seconds": 0.7940784490019723,
+      "bytes": 968621,
       "pages": 29,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 1.0165509960006602,
-      "bytes": 11776139,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.324730037998961,
+      "bytes": 13434552,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 1.1994448559999,
-      "bytes": 10381245,
-      "pages": 35,
+      "seconds": 1.516792244001408,
+      "bytes": 10349687,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.1291984180006693,
-      "bytes": 2752026,
+      "seconds": 1.0428093720001925,
+      "bytes": 2636885,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8884318259988504,
-      "bytes": 6185512,
-      "pages": 36,
+      "seconds": 1.006605718001083,
+      "bytes": 6127707,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.246648587000891,
-      "bytes": 6489173,
-      "pages": 30,
+      "seconds": 1.116981245999341,
+      "bytes": 6115742,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.009036264000315,
+      "bytes": 6506928,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0390239320004184,
-      "bytes": 6087872,
+      "seconds": 1.1265610020018357,
+      "bytes": 11819263,
       "pages": 35,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.244901927000683,
-      "bytes": 11604745,
-      "pages": 30,
+      "seconds": 1.0389532609988237,
+      "bytes": 7773400,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.1130494770004589,
-      "bytes": 2612508,
+      "seconds": 0.9040890260002925,
+      "bytes": 1006858,
+      "pages": 30,
+      "shards": 1
+    },
+    {
+      "seconds": 1.4120586079989152,
+      "bytes": 10016675,
       "pages": 37,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0292591729994456,
+      "bytes": 2310644,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9836850789997698,
-      "bytes": 10345276,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.012728962003166,
+      "bytes": 6259160,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1697606620000442,
-      "bytes": 7536660,
-      "pages": 38,
-      "shards": 3
+      "seconds": 0.9536746940029843,
+      "bytes": 6477995,
+      "pages": 23,
+      "shards": 2
     },
     {
-      "seconds": 0.8494176380008867,
-      "bytes": 947075,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.2943796040017332,
+      "bytes": 12517532,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.246135413999582,
-      "bytes": 6441176,
+      "seconds": 1.182365384996956,
+      "bytes": 6177943,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9687062859993603,
-      "bytes": 7942604,
-      "pages": 25,
-      "shards": 3
+      "seconds": 0.8839036259996647,
+      "bytes": 2483088,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.493546093999612,
-      "bytes": 10942153,
-      "pages": 44,
+      "seconds": 1.165388396999333,
+      "bytes": 6302666,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.185240850001719,
-      "bytes": 5667842,
-      "pages": 28,
+      "seconds": 1.2500933689989324,
+      "bytes": 4989357,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 0.9565514620007889,
-      "bytes": 2665835,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.2095713759990758,
+      "bytes": 7770394,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.161401912002475,
-      "bytes": 5953132,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.1173389949981356,
+      "bytes": 12086613,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.9063561590010067,
-      "bytes": 11430108,
-      "pages": 37,
+      "seconds": 0.9005163840010937,
+      "bytes": 1099069,
+      "pages": 30,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1803624109998054,
+      "bytes": 11418078,
+      "pages": 40,
       "shards": 4
+    },
+    {
+      "seconds": 1.0558311360000516,
+      "bytes": 6077678,
+      "pages": 30,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:442d5a63837b6ac40b1f2659fb1a8dca8d13b6c8efafd35b7dd88a3f95501c4e
-size 721775615
+oid sha256:b8e924ba03c8c90f3a051947fd0eb5994e5b786f87c64a7022243d5bfe8f0021
+size 721769151