Qwen90M step 145000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 140000,
+  "step": 145000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "1f6a78c79a1798f6066bb26ce17abbc1d73d28136a3d86664a0bb32df93cffe4",
+    "model.safetensors": "f82ef0b41a9f5a444a6d4c213cf42228dda7b7856086639bd309c8f1264486c9",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "ac8f437b95021c6471b19562d50262039a426962f6645567d059bda2e6caecb2",
+    "trainer.pt": "81681141f6dfe127e198e16da88d8c38af13ada9c9f0a4963e0cd992eb0b365c",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "10fa55d76d1a4ec4f9132ef494d950eba8aa9e6cecd5fa1ceb70ba6de35537e4",
-    "shard_scheduler.json": "d0f4806bcb4805a31116b9a978bfd03cd9d8a2f854a359c44e04747555b4537f",
-    "coverage.json": "4066b47af8c393c7a151754ca159ed1759530fda15d4f6c2c2fa2b7dd82c464c",
+    "shard_scheduler.json": "33c9db18de388458070d6df886e5bda2cedf9dabd3e716f543d2e756f18de9cc",
+    "coverage.json": "088c76466a16e8303fb88ede838eb2896c816dbf26f02d114c54169dacd144e5",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 140000,
+  "step": 145000,
   "stats": {
-    "climbmix/blocks": 5040000,
-    "climbmix/targets": 8422837629,
-    "climbmix/input_tokens": 8423371884,
-    "ultra_qa/blocks": 1890000,
-    "ultra_qa/targets": 3082816374,
-    "ultra_qa/input_tokens": 3082822297,
-    "cortex/blocks": 1890000,
-    "cortex/targets": 71865436,
-    "cortex/input_tokens": 3004020105,
-    "ultra_style/blocks": 1890000,
-    "ultra_style/targets": 2989849456,
-    "ultra_style/input_tokens": 2990006029,
-    "openmath/blocks": 1260000,
-    "openmath/targets": 1946691423,
-    "openmath/input_tokens": 1947340330,
-    "rewrite6/blocks": 630000,
-    "rewrite6/targets": 1032782702,
-    "rewrite6/input_tokens": 1033915120
+    "climbmix/blocks": 5220000,
+    "climbmix/targets": 8723475320,
+    "climbmix/input_tokens": 8724028649,
+    "ultra_qa/blocks": 1957500,
+    "ultra_qa/targets": 3193055973,
+    "ultra_qa/input_tokens": 3193062132,
+    "cortex/blocks": 1957500,
+    "cortex/targets": 74442843,
+    "cortex/input_tokens": 3111300467,
+    "ultra_style/blocks": 1957500,
+    "ultra_style/targets": 3096311658,
+    "ultra_style/input_tokens": 3096473872,
+    "openmath/blocks": 1305000,
+    "openmath/targets": 2016316779,
+    "openmath/input_tokens": 2016988997,
+    "rewrite6/blocks": 652500,
+    "rewrite6/targets": 1069684284,
+    "rewrite6/input_tokens": 1070857144
   },
   "sources": {
     "climbmix": {
-      "page": 690,
-      "row": 3762,
+      "page": 715,
+      "row": 2681,
       "start": {
-        "file": 33,
+        "file": 34,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 27,
+          "row_group": 44,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 2830002
+      "offset": 0,
+      "documents": 2931321
     },
     "rewrite6": {
-      "page": 42,
-      "row": 3438,
+      "page": 6,
+      "row": 3005,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 172032
+          "row": 24576
         }
       },
-      "epoch": 7,
-      "offset": 0,
-      "documents": 1613298
+      "epoch": 8,
+      "offset": 2048,
+      "documents": 1670813
     },
     "openmath": {
-      "page": 431,
-      "row": 1976,
+      "page": 446,
+      "row": 3868,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 18,
-          "row": 776
+          "row_group": 80,
+          "row": 216
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 1767352
+      "documents": 1830684
     },
     "ultra_qa": {
-      "page": 186,
-      "row": 2242,
+      "page": 193,
+      "row": 848,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 48427
+          "row_group": 4,
+          "row": 12512
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 764098
+      "offset": 0,
+      "documents": 791376
     },
     "ultra_style": {
-      "page": 284,
-      "row": 2574,
+      "page": 294,
+      "row": 3383,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 5,
-          "row": 52597
+          "row_group": 6,
+          "row": 9157
         }
       },
       "epoch": 0,
-      "offset": 6144,
-      "documents": 1165838
+      "offset": 2048,
+      "documents": 1207607
     },
     "cortex": {
-      "page": 36653,
-      "row": 34,
+      "page": 37962,
+      "row": 8,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 4691618
+      "documents": 4859144
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 140000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "1f6a78c79a1798f6066bb26ce17abbc1d73d28136a3d86664a0bb32df93cffe4", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "ac8f437b95021c6471b19562d50262039a426962f6645567d059bda2e6caecb2", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "10fa55d76d1a4ec4f9132ef494d950eba8aa9e6cecd5fa1ceb70ba6de35537e4", "shard_scheduler.json": "d0f4806bcb4805a31116b9a978bfd03cd9d8a2f854a359c44e04747555b4537f", "coverage.json": "4066b47af8c393c7a151754ca159ed1759530fda15d4f6c2c2fa2b7dd82c464c", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000140000"}
+{"step": 145000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "f82ef0b41a9f5a444a6d4c213cf42228dda7b7856086639bd309c8f1264486c9", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "81681141f6dfe127e198e16da88d8c38af13ada9c9f0a4963e0cd992eb0b365c", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "10fa55d76d1a4ec4f9132ef494d950eba8aa9e6cecd5fa1ceb70ba6de35537e4", "shard_scheduler.json": "33c9db18de388458070d6df886e5bda2cedf9dabd3e716f543d2e756f18de9cc", "coverage.json": "088c76466a16e8303fb88ede838eb2896c816dbf26f02d114c54169dacd144e5", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000145000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1f6a78c79a1798f6066bb26ce17abbc1d73d28136a3d86664a0bb32df93cffe4
+oid sha256:f82ef0b41a9f5a444a6d4c213cf42228dda7b7856086639bd309c8f1264486c9
 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": 75954.54383687535,
-      "required_pages_per_second_upper_estimate": 0.0065835473547489675
+      "required_token_positions_per_second": 75819.42597709605,
+      "required_pages_per_second_upper_estimate": 0.006571835681116382
     },
     {
       "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": 9494.31797960942,
-      "required_pages_per_second_upper_estimate": 0.0037874253947699933
+      "required_token_positions_per_second": 9477.428247137006,
+      "required_pages_per_second_upper_estimate": 0.003780687827962744
     },
     {
       "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": 18988.63595921884,
-      "required_pages_per_second_upper_estimate": 0.004392686740586353
+      "required_token_positions_per_second": 18954.856494274012,
+      "required_pages_per_second_upper_estimate": 0.0043848724558696464
     },
     {
       "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": 28482.953938828257,
-      "required_pages_per_second_upper_estimate": 0.0017344624563645405
+      "required_token_positions_per_second": 28432.28474141102,
+      "required_pages_per_second_upper_estimate": 0.0017313769680825642
     },
     {
       "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": 28482.953938828257,
-      "required_pages_per_second_upper_estimate": 0.0021926966400738616
+      "required_token_positions_per_second": 28432.28474141102,
+      "required_pages_per_second_upper_estimate": 0.002188795984995367
     },
     {
       "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": 28482.953938828257,
-      "required_pages_per_second_upper_estimate": 0.3536629616055758
+      "required_token_positions_per_second": 28432.28474141102,
+      "required_pages_per_second_upper_estimate": 0.35303381975254877
     }
   ],
   "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": 688946,
-    "producer_batch_seconds": 0.1653632896328645,
-    "producer_p95_seconds": 0.35368149500209256,
-    "gpu_update_seconds": 0.9706858375496638,
-    "producer_batches_per_second": 6.0472914043992185,
-    "gpu_batches_per_second": 1.0301994335513691,
-    "producer_headroom": 5.870020121786138,
+    "producer_batch_seconds": 0.22319741781251423,
+    "producer_p95_seconds": 0.37801911600035965,
+    "gpu_update_seconds": 0.9724156975584616,
+    "producer_batches_per_second": 4.4803385711209245,
+    "gpu_batches_per_second": 1.0283667802883036,
+    "producer_headroom": 4.3567515569346345,
     "sustainable": true,
-    "buffer_seconds": 60.182521928079154,
+    "buffer_seconds": 60.28977324862462,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 80,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 287573.6403975973,
-    "upload_bytes_per_second": 4002366.2795159263,
+    "generation_bytes_per_second": 549008.0145004862,
+    "upload_bytes_per_second": 5070479.625359649,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 61.706608857592954,
+    "estimated_full_shards_per_hour": 117.80433965931834,
     "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": 1.1858630629976687,
-      "bytes": 13358845,
-      "pages": 34,
-      "shards": 4
-    },
-    {
-      "seconds": 0.9764441150000494,
-      "bytes": 4826687,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.8942368999996688,
+      "bytes": 850966,
+      "pages": 25,
+      "shards": 1
     },
     {
-      "seconds": 1.4679961010006082,
-      "bytes": 7426505,
+      "seconds": 1.4387539480012492,
+      "bytes": 11588272,
       "pages": 38,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 1.038269829001365,
-      "bytes": 1088045,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.6636693889995513,
+      "bytes": 11864781,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.3681885359983426,
-      "bytes": 5869873,
-      "pages": 29,
+      "seconds": 1.1722596879990306,
+      "bytes": 1208158,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.2775829050005996,
-      "bytes": 12210679,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.2069239850025042,
+      "bytes": 7251816,
+      "pages": 27,
+      "shards": 3
     },
     {
-      "seconds": 1.1223962459989707,
-      "bytes": 6954584,
-      "pages": 39,
+      "seconds": 1.169119137000962,
+      "bytes": 6109087,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.6785005409983569,
-      "bytes": 5701009,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.4141185940025025,
+      "bytes": 7704490,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.9818456940010947,
-      "bytes": 2688719,
-      "pages": 36,
+      "seconds": 1.2197817560008843,
+      "bytes": 5784149,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0817985579997185,
-      "bytes": 6075598,
-      "pages": 20,
+      "seconds": 1.3641967180010397,
+      "bytes": 7009240,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3699396060001163,
-      "bytes": 12184416,
-      "pages": 47,
+      "seconds": 1.4704428500008362,
+      "bytes": 11028707,
+      "pages": 30,
       "shards": 4
     },
     {
-      "seconds": 0.945849695999641,
-      "bytes": 1129028,
-      "pages": 33,
+      "seconds": 0.9246157949964982,
+      "bytes": 1204930,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.1423426910005219,
-      "bytes": 5919036,
-      "pages": 28,
+      "seconds": 1.0427682160006952,
+      "bytes": 5985007,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1784255179991305,
-      "bytes": 13145531,
-      "pages": 35,
-      "shards": 5
+      "seconds": 1.1093607039983908,
+      "bytes": 7338185,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.339017384001636,
-      "bytes": 6203121,
+      "seconds": 0.9098767819996283,
+      "bytes": 5901703,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.3240333100002317,
-      "bytes": 6490741,
+      "seconds": 1.1044195929971465,
+      "bytes": 13283238,
+      "pages": 34,
+      "shards": 4
+    },
+    {
+      "seconds": 0.8937192140001571,
+      "bytes": 6207827,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9587836459977552,
-      "bytes": 2593270,
+      "seconds": 0.9678468759993848,
+      "bytes": 1151038,
       "pages": 35,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 0.8898652169991692,
-      "bytes": 5769808,
-      "pages": 27,
+      "seconds": 1.3402203920013562,
+      "bytes": 12178587,
+      "pages": 33,
+      "shards": 4
+    },
+    {
+      "seconds": 0.8894950190006057,
+      "bytes": 1180133,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1408973270008573,
+      "bytes": 5952533,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 2.3888570920025813,
-      "bytes": 6457509,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.4523514239990618,
+      "bytes": 13500548,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 1.143103587000951,
-      "bytes": 11836705,
-      "pages": 33,
-      "shards": 3
+      "seconds": 0.8281688100032625,
+      "bytes": 661491,
+      "pages": 19,
+      "shards": 1
     },
     {
-      "seconds": 1.1624738930004241,
-      "bytes": 6035324,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.42470605500057,
+      "bytes": 11031230,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.368994181000744,
-      "bytes": 7247693,
-      "pages": 32,
-      "shards": 3
+      "seconds": 0.8167821940005524,
+      "bytes": 1161839,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 0.8721044880003319,
-      "bytes": 6235096,
-      "pages": 36,
+      "seconds": 0.9240367780002998,
+      "bytes": 5199805,
+      "pages": 10,
       "shards": 2
     },
     {
-      "seconds": 1.1694963700028893,
-      "bytes": 895144,
-      "pages": 30,
+      "seconds": 1.4070672229972843,
+      "bytes": 1010316,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.171581717000663,
-      "bytes": 7061925,
-      "pages": 24,
-      "shards": 3
+      "seconds": 1.181387053002254,
+      "bytes": 11521167,
+      "pages": 38,
+      "shards": 4
     },
     {
-      "seconds": 1.0121167409997724,
-      "bytes": 6826316,
-      "pages": 33,
+      "seconds": 1.134901017001539,
+      "bytes": 6770130,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.0728631890015095,
-      "bytes": 9404752,
+      "seconds": 1.1226096480022534,
+      "bytes": 7935280,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 32.4843381790015,
-      "bytes": 7401467,
-      "pages": 32,
-      "shards": 3
+      "seconds": 0.982757900001161,
+      "bytes": 5911177,
+      "pages": 25,
+      "shards": 2
     },
     {
-      "seconds": 1.1189806990005309,
-      "bytes": 7079866,
-      "pages": 44,
+      "seconds": 1.134857662000286,
+      "bytes": 5917855,
+      "pages": 41,
       "shards": 2
     },
     {
-      "seconds": 1.104166561999591,
-      "bytes": 7255454,
-      "pages": 33,
+      "seconds": 1.5188524260011036,
+      "bytes": 7655478,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.0669172650013934,
-      "bytes": 6210420,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.7887198950011225,
+      "bytes": 905384,
+      "pages": 25,
+      "shards": 1
     },
     {
-      "seconds": 1.0871770170015225,
-      "bytes": 6926231,
-      "pages": 33,
-      "shards": 2
+      "seconds": 4.007056702997943,
+      "bytes": 13411793,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.2097226180012512,
-      "bytes": 7368748,
-      "pages": 29,
+      "seconds": 1.3467935920016316,
+      "bytes": 10736953,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1227261329986504,
-      "bytes": 5278941,
-      "pages": 36,
+      "seconds": 0.8973097269990831,
+      "bytes": 792959,
+      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 1.216369551999378,
-      "bytes": 7462357,
-      "pages": 34,
+      "seconds": 1.2533608519988775,
+      "bytes": 7837535,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.5228881869988982,
-      "bytes": 5895199,
+      "seconds": 1.1466966719999618,
+      "bytes": 6019789,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0030821439977444,
-      "bytes": 1174580,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.8239342810011294,
-      "bytes": 17807257,
+      "seconds": 1.0688578599983884,
+      "bytes": 4660632,
       "pages": 31,
-      "shards": 5
+      "shards": 2
     },
     {
-      "seconds": 1.1584871390004992,
-      "bytes": 5888359,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.4090757320009288,
+      "bytes": 13354980,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 0.8503742749999219,
-      "bytes": 1173601,
-      "pages": 34,
+      "seconds": 1.0085502450019703,
+      "bytes": 1017261,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.4903823460008425,
-      "bytes": 7681898,
-      "pages": 39,
+      "seconds": 1.7747774790004769,
+      "bytes": 7374031,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.8942368999996688,
-      "bytes": 850966,
-      "pages": 25,
-      "shards": 1
+      "seconds": 0.9477112800013856,
+      "bytes": 5792070,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.4387539480012492,
-      "bytes": 11588272,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.0662493220006581,
+      "bytes": 5942385,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.6636693889995513,
-      "bytes": 11864781,
-      "pages": 37,
+      "seconds": 1.245676413000183,
+      "bytes": 7693974,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1722596879990306,
-      "bytes": 1208158,
-      "pages": 35,
+      "seconds": 1.0202759780004271,
+      "bytes": 6615812,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2069239850025042,
-      "bytes": 7251816,
-      "pages": 27,
+      "seconds": 1.4830753700007335,
+      "bytes": 9974834,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.169119137000962,
-      "bytes": 6109087,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.272736505001376,
+      "bytes": 7611348,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.4141185940025025,
-      "bytes": 7704490,
-      "pages": 38,
+      "seconds": 0.8344998959983059,
+      "bytes": 1093697,
+      "pages": 33,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3662152600008994,
+      "bytes": 7153184,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.2197817560008843,
-      "bytes": 5784149,
-      "pages": 28,
+      "seconds": 1.6717624239972793,
+      "bytes": 5422811,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.3641967180010397,
-      "bytes": 7009240,
+      "seconds": 1.1521110989997396,
+      "bytes": 12146138,
       "pages": 38,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.4704428500008362,
-      "bytes": 11028707,
+      "seconds": 1.1019259579989011,
+      "bytes": 7338055,
       "pages": 30,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 0.9246157949964982,
-      "bytes": 1204930,
-      "pages": 34,
+      "seconds": 0.7986074580003333,
+      "bytes": 1164850,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.0427682160006952,
-      "bytes": 5985007,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1093607039983908,
-      "bytes": 7338185,
-      "pages": 30,
+      "seconds": 1.6704007100015588,
+      "bytes": 10030549,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 0.9098767819996283,
-      "bytes": 5901703,
+      "seconds": 2.951283398000669,
+      "bytes": 2617533,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1044195929971465,
-      "bytes": 13283238,
-      "pages": 34,
-      "shards": 4
-    },
-    {
-      "seconds": 0.8937192140001571,
-      "bytes": 6207827,
+      "seconds": 1.1606040469996515,
+      "bytes": 6236824,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9678468759993848,
-      "bytes": 1151038,
-      "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 1.3402203920013562,
-      "bytes": 12178587,
-      "pages": 33,
-      "shards": 4
+      "seconds": 4.655135680000967,
+      "bytes": 7374926,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 0.8894950190006057,
-      "bytes": 1180133,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.311550709000585,
+      "bytes": 11722471,
+      "pages": 42,
+      "shards": 3
     },
     {
-      "seconds": 1.1408973270008573,
-      "bytes": 5952533,
+      "seconds": 0.7739134580006066,
+      "bytes": 5816616,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.4523514239990618,
-      "bytes": 13500548,
-      "pages": 42,
-      "shards": 4
+      "seconds": 1.117633730002126,
+      "bytes": 7565346,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 0.8281688100032625,
-      "bytes": 661491,
-      "pages": 19,
+      "seconds": 0.8082868439996673,
+      "bytes": 1195987,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.42470605500057,
-      "bytes": 11031230,
-      "pages": 41,
+      "seconds": 1.157742789000622,
+      "bytes": 11040923,
+      "pages": 34,
       "shards": 4
+    },
+    {
+      "seconds": 0.8454459890017461,
+      "bytes": 981466,
+      "pages": 29,
+      "shards": 1
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ac8f437b95021c6471b19562d50262039a426962f6645567d059bda2e6caecb2
-size 721768575
+oid sha256:81681141f6dfe127e198e16da88d8c38af13ada9c9f0a4963e0cd992eb0b365c
+size 721762943