Qwen90M step 500000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 495000,
+  "step": 500000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "04a9c9ab7806b229ff0c1f81bf062a0cb0f4ff40115f9dc54fcd35653143dba5",
+    "model.safetensors": "b32b5405757b9ffe78293bedfdb956dae40e8488a1c520be2cde0c5aed95c84e",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "b09148f8bdb03fa1cdc8767b6fe281fcc58f0012b4c4869a74f3162a6d4c4a19",
+    "trainer.pt": "442d5a63837b6ac40b1f2659fb1a8dca8d13b6c8efafd35b7dd88a3f95501c4e",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "234744063e094dfd7b525da4e181ac493fac400c830f871c7e8c59127ed10dc9",
-    "shard_scheduler.json": "5090f346773104837959c2e40fec42df5511a51ba2bc53922efca54c043ec515",
-    "coverage.json": "79dfd44233cf538b7f2dfa46bc0fdc879f8607366262089a9f9c1e5fcd1d4480",
+    "shard_scheduler.json": "28346349a8e1d8ef3ce4b1c69a6243a3f826cee646b53fd277baf701c2210e6e",
+    "coverage.json": "19aa4f8d54332a074a778e21560e24dcd35db2f280ad9c717dbc173725ed6762",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 495000,
+  "step": 500000,
   "stats": {
-    "climbmix/blocks": 17820000,
-    "climbmix/targets": 29786461296,
-    "climbmix/input_tokens": 29788349282,
-    "ultra_qa/blocks": 6682500,
-    "ultra_qa/targets": 10901546468,
-    "ultra_qa/input_tokens": 10901567474,
-    "cortex/blocks": 6682500,
-    "cortex/targets": 254162427,
-    "cortex/input_tokens": 10622791425,
-    "ultra_style/blocks": 6682500,
-    "ultra_style/targets": 10570465550,
-    "ultra_style/input_tokens": 10571019587,
-    "openmath/blocks": 4455000,
-    "openmath/targets": 6883812928,
-    "openmath/input_tokens": 6886106549,
-    "rewrite6/blocks": 2227500,
-    "rewrite6/targets": 3651663877,
-    "rewrite6/input_tokens": 3655667526
+    "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
   },
   "sources": {
     "climbmix": {
-      "page": 2441,
-      "row": 677,
+      "page": 2465,
+      "row": 3704,
       "start": {
-        "file": 118,
+        "file": 119,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 9,
+          "row_group": 22,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 9999013
+      "offset": 0,
+      "documents": 10100344
     },
     "rewrite6": {
-      "page": 38,
-      "row": 2269,
+      "page": 2,
+      "row": 1645,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 155648
+          "row": 8192
         }
       },
-      "epoch": 27,
+      "epoch": 28,
       "offset": 0,
-      "documents": 5703825
+      "documents": 5761149
     },
     "openmath": {
-      "page": 1525,
-      "row": 1171,
+      "page": 1540,
+      "row": 2675,
       "start": {
         "file": 14,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 133,
-          "row": 300
+          "row_group": 194,
+          "row": 740
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 6247571
+      "offset": 0,
+      "documents": 6310515
     },
     "ultra_qa": {
-      "page": 659,
-      "row": 2052,
+      "page": 666,
+      "row": 700,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 1,
-          "row": 36676
+          "row_group": 2,
+          "row": 775
         }
       },
       "epoch": 0,
       "offset": 4096,
-      "documents": 2701316
+      "documents": 2728636
     },
     "ultra_style": {
-      "page": 1006,
-      "row": 1193,
+      "page": 1016,
+      "row": 2694,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 0,
-          "row": 10924
+          "row": 51884
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 4121769
+      "offset": 4096,
+      "documents": 4164230
     },
     "cortex": {
-      "page": 129606,
-      "row": 60,
+      "page": 130914,
+      "row": 26,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 16589628
+      "documents": 16757018
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 495000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "04a9c9ab7806b229ff0c1f81bf062a0cb0f4ff40115f9dc54fcd35653143dba5", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "b09148f8bdb03fa1cdc8767b6fe281fcc58f0012b4c4869a74f3162a6d4c4a19", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "234744063e094dfd7b525da4e181ac493fac400c830f871c7e8c59127ed10dc9", "shard_scheduler.json": "5090f346773104837959c2e40fec42df5511a51ba2bc53922efca54c043ec515", "coverage.json": "79dfd44233cf538b7f2dfa46bc0fdc879f8607366262089a9f9c1e5fcd1d4480", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000495000"}
+{"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"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:04a9c9ab7806b229ff0c1f81bf062a0cb0f4ff40115f9dc54fcd35653143dba5
+oid sha256:b32b5405757b9ffe78293bedfdb956dae40e8488a1c520be2cde0c5aed95c84e
 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": 77246.17314450147,
-      "required_pages_per_second_upper_estimate": 0.005344069493952184
+      "required_token_positions_per_second": 76297.59507963236,
+      "required_pages_per_second_upper_estimate": 0.005278444662420188
     },
     {
       "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": 9655.771643062684,
-      "required_pages_per_second_upper_estimate": 0.004042669951509708
+      "required_token_positions_per_second": 9537.199384954045,
+      "required_pages_per_second_upper_estimate": 0.0039930262231099335
     },
     {
       "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": 19311.54328612537,
-      "required_pages_per_second_upper_estimate": 0.0040220851868558866
+      "required_token_positions_per_second": 19074.39876990809,
+      "required_pages_per_second_upper_estimate": 0.003972694238049278
     },
     {
       "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": 28967.31492918805,
-      "required_pages_per_second_upper_estimate": 0.0017365462994759575
+      "required_token_positions_per_second": 28611.598154862135,
+      "required_pages_per_second_upper_estimate": 0.0017152216220031838
     },
     {
       "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": 28967.31492918805,
-      "required_pages_per_second_upper_estimate": 0.002252081321991498
+      "required_token_positions_per_second": 28611.598154862135,
+      "required_pages_per_second_upper_estimate": 0.0022244259074203927
     },
     {
       "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": 28967.31492918805,
-      "required_pages_per_second_upper_estimate": 0.3715616132321039
+      "required_token_positions_per_second": 28611.598154862135,
+      "required_pages_per_second_upper_estimate": 0.36699886038996593
     }
   ],
   "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": 0.18283066840626816,
-    "producer_p95_seconds": 0.22040481899966835,
-    "gpu_update_seconds": 0.9544550493405005,
-    "producer_batches_per_second": 5.469541892052264,
-    "gpu_batches_per_second": 1.0477182772420448,
-    "producer_headroom": 5.220431876448678,
+    "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,
     "sustainable": true,
-    "buffer_seconds": 60.13066810845153,
+    "buffer_seconds": 60.87824911325346,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 78,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 476173.2566712523,
-    "upload_bytes_per_second": 5837253.21659938,
+    "generation_bytes_per_second": 321426.0121983039,
+    "upload_bytes_per_second": 5933281.998072939,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 102.17569613555123,
+    "estimated_full_shards_per_hour": 68.970539803141,
     "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": 1.011828892000267,
-      "bytes": 8292186,
-      "pages": 35,
+      "seconds": 1.1922360469998239,
+      "bytes": 10997251,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.0172779360000277,
-      "bytes": 6155138,
-      "pages": 37,
+      "seconds": 1.0726386229998752,
+      "bytes": 2419938,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1509364789999381,
-      "bytes": 12144311,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.240178467000078,
+      "bytes": 6103593,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.977909455000372,
-      "bytes": 856165,
-      "pages": 26,
-      "shards": 1
+      "seconds": 0.8772162379991642,
+      "bytes": 2610307,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.183507477000603,
+      "bytes": 15672429,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 1.0899097739993522,
-      "bytes": 6114737,
+      "seconds": 1.1264393319997907,
+      "bytes": 6331952,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.358950516000732,
-      "bytes": 2583429,
-      "pages": 37,
+      "seconds": 0.856032101000892,
+      "bytes": 2311911,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.304920979000599,
-      "bytes": 11842449,
-      "pages": 27,
+      "seconds": 1.5313736190000782,
+      "bytes": 6009990,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.2498191829999996,
-      "bytes": 5414761,
-      "pages": 37,
+      "seconds": 1.0109308320006676,
+      "bytes": 5084502,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0708195489996797,
-      "bytes": 7566500,
-      "pages": 34,
+      "seconds": 1.1290311129996553,
+      "bytes": 7603249,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1241090929997881,
-      "bytes": 6087377,
+      "seconds": 1.4725222399993072,
+      "bytes": 12055069,
       "pages": 36,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.0640653479995308,
-      "bytes": 6920167,
-      "pages": 38,
+      "seconds": 1.2178526319985394,
+      "bytes": 7463726,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 0.9458256010002515,
-      "bytes": 5937789,
-      "pages": 27,
+      "seconds": 1.3964880730000004,
+      "bytes": 6495644,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3173817269998835,
-      "bytes": 11936639,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.2645150809985353,
+      "bytes": 6000960,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 0.9754752660001031,
-      "bytes": 2583341,
-      "pages": 36,
+      "seconds": 0.9461197429991444,
+      "bytes": 2470605,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.5200459669995325,
-      "bytes": 9596109,
-      "pages": 30,
-      "shards": 4
+      "seconds": 1.2941537009992317,
+      "bytes": 6225550,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1015423869994265,
-      "bytes": 7651468,
-      "pages": 38,
+      "seconds": 1.2335469089994149,
+      "bytes": 11703737,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.9403381189995343,
-      "bytes": 1076403,
+      "seconds": 1.1459508749994711,
+      "bytes": 7841640,
       "pages": 34,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 1.025973077999879,
-      "bytes": 6061719,
-      "pages": 37,
+      "seconds": 1.0024716899988562,
+      "bytes": 5917383,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.228285135999613,
-      "bytes": 12817063,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.1789036200007104,
+      "bytes": 2313669,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 0.9785899170001358,
-      "bytes": 5669538,
-      "pages": 27,
+      "seconds": 1.5398991070014745,
+      "bytes": 10300581,
+      "pages": 40,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9979469910012995,
+      "bytes": 6506963,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2632895730002929,
-      "bytes": 7532013,
-      "pages": 38,
+      "seconds": 1.1795985310000106,
+      "bytes": 8204299,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.3679505390000486,
-      "bytes": 1187716,
-      "pages": 35,
-      "shards": 1
+      "seconds": 0.9598508790004416,
+      "bytes": 6041026,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.3033395869997548,
-      "bytes": 10049773,
-      "pages": 30,
+      "seconds": 0.9930732779994287,
+      "bytes": 6062259,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 0.8627702979993046,
-      "bytes": 2455057,
+      "seconds": 1.0064647640010662,
+      "bytes": 5698061,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2129529189996902,
-      "bytes": 11955081,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.1021013760000642,
+      "bytes": 5987765,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.95688545899975,
-      "bytes": 1097352,
+      "seconds": 1.0322518390003097,
+      "bytes": 2308031,
       "pages": 30,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.40065100099946,
-      "bytes": 12237786,
-      "pages": 38,
+      "seconds": 1.3352466599990294,
+      "bytes": 15653746,
+      "pages": 42,
       "shards": 4
     },
     {
-      "seconds": 1.0810250989998167,
-      "bytes": 5909598,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1028008130015223,
+      "bytes": 1043427,
+      "pages": 31,
+      "shards": 1
     },
     {
-      "seconds": 0.9096889889988233,
-      "bytes": 1202765,
+      "seconds": 0.9759294559989939,
+      "bytes": 7445734,
       "pages": 34,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 1.2456794589998026,
-      "bytes": 7629808,
-      "pages": 38,
+      "seconds": 1.0883692229999724,
+      "bytes": 10848676,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.9217846150004334,
-      "bytes": 6311419,
-      "pages": 21,
+      "seconds": 1.0698365700009163,
+      "bytes": 2697855,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.1030007570007,
-      "bytes": 11547889,
-      "pages": 39,
-      "shards": 4
-    },
-    {
-      "seconds": 1.05019463699864,
-      "bytes": 6120696,
-      "pages": 37,
+      "seconds": 0.9394886710015271,
+      "bytes": 6144198,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.070229210999969,
-      "bytes": 1151046,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.0910024349996092,
+      "bytes": 11590114,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.0475356750011997,
-      "bytes": 11473932,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.1193175999997038,
+      "bytes": 7119189,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.2914608989995031,
-      "bytes": 6033641,
+      "seconds": 1.3181563210000604,
+      "bytes": 6116642,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.115209474999574,
-      "bytes": 812365,
-      "pages": 23,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1693048819997784,
-      "bytes": 7948420,
-      "pages": 20,
+      "seconds": 0.95897385599892,
+      "bytes": 2279278,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.2017747059999238,
-      "bytes": 5823273,
-      "pages": 30,
+      "seconds": 1.0730731740004558,
+      "bytes": 6315193,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1922360469998239,
-      "bytes": 10997251,
-      "pages": 39,
+      "seconds": 1.276775304999319,
+      "bytes": 10081435,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.0726386229998752,
-      "bytes": 2419938,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1636577470017073,
+      "bytes": 8230638,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.240178467000078,
-      "bytes": 6103593,
-      "pages": 36,
+      "seconds": 1.2417163159989286,
+      "bytes": 6099161,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8772162379991642,
-      "bytes": 2610307,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.1390252959990903,
+      "bytes": 7728259,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.183507477000603,
-      "bytes": 15672429,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.331549176999033,
+      "bytes": 5708517,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "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,
+      "seconds": 0.8494176380008867,
+      "bytes": 947075,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.246135413999582,
+      "bytes": 6441176,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1789036200007104,
-      "bytes": 2313669,
-      "pages": 29,
-      "shards": 2
+      "seconds": 0.9687062859993603,
+      "bytes": 7942604,
+      "pages": 25,
+      "shards": 3
     },
     {
-      "seconds": 1.5398991070014745,
-      "bytes": 10300581,
-      "pages": 40,
+      "seconds": 1.493546093999612,
+      "bytes": 10942153,
+      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 0.9979469910012995,
-      "bytes": 6506963,
-      "pages": 36,
+      "seconds": 1.185240850001719,
+      "bytes": 5667842,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1795985310000106,
-      "bytes": 8204299,
-      "pages": 31,
-      "shards": 3
+      "seconds": 0.9565514620007889,
+      "bytes": 2665835,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.9598508790004416,
-      "bytes": 6041026,
-      "pages": 37,
+      "seconds": 1.161401912002475,
+      "bytes": 5953132,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 0.9930732779994287,
-      "bytes": 6062259,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.9063561590010067,
+      "bytes": 11430108,
+      "pages": 37,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b09148f8bdb03fa1cdc8767b6fe281fcc58f0012b4c4869a74f3162a6d4c4a19
-size 721760383
+oid sha256:442d5a63837b6ac40b1f2659fb1a8dca8d13b6c8efafd35b7dd88a3f95501c4e
+size 721775615