Qwen90M step 380000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 375000,
+  "step": 380000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "6610966b863295dd906696faac88a04c6ff0496a879a152f0273bb1b9d396255",
+    "model.safetensors": "c1d74f83e24ba5ed4a98e7a8650879fde80d5e46dcea28dfeb99c4ff2c71c650",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "406bc235a118a90d6ce9168e0c61f5aeee892c2cd7892ea2dfe941963fb47bc3",
+    "trainer.pt": "96586ee6a38bb3d010196afc962d100dcda3fae0e33facf3c5450cee743fa5c6",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "e3923fa86352f6d878fd8d37b6b9fc374577f52be522203b0b865badf630da15",
-    "shard_scheduler.json": "3446bd4676a0432db7c222c8eed67dd16a89c65bbd51ec69a9afb2a131cff95e",
-    "coverage.json": "eb5f67654bace9fef82be9a9d8a490ab08cd158a627ef4268f68ce6a57faf403",
+    "shard_scheduler.json": "99b8b50757cc145af0cc487dcd657bdacdcccce20a0ed8fcd882dd2d990666a8",
+    "coverage.json": "1bcb49c8690d4fac9adfe028153b744eff8144e6c17080d648fa5eaf6ebf1532",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 375000,
+  "step": 380000,
   "stats": {
-    "climbmix/blocks": 13500000,
-    "climbmix/targets": 22566571953,
-    "climbmix/input_tokens": 22568001890,
-    "ultra_qa/blocks": 5062500,
-    "ultra_qa/targets": 8258873649,
-    "ultra_qa/input_tokens": 8258889605,
-    "cortex/blocks": 5062500,
-    "cortex/targets": 192584950,
-    "cortex/input_tokens": 8047684352,
-    "ultra_style/blocks": 5062500,
-    "ultra_style/targets": 8007585321,
-    "ultra_style/input_tokens": 8008005686,
-    "openmath/blocks": 3375000,
-    "openmath/targets": 5215242560,
-    "openmath/input_tokens": 5216980495,
-    "rewrite6/blocks": 1687500,
-    "rewrite6/targets": 2766400210,
-    "rewrite6/input_tokens": 2769433064
+    "climbmix/blocks": 13680000,
+    "climbmix/targets": 22867425170,
+    "climbmix/input_tokens": 22868874051,
+    "ultra_qa/blocks": 5130000,
+    "ultra_qa/targets": 8368921036,
+    "ultra_qa/input_tokens": 8368937191,
+    "cortex/blocks": 5130000,
+    "cortex/targets": 195140335,
+    "cortex/input_tokens": 8155034719,
+    "ultra_style/blocks": 5130000,
+    "ultra_style/targets": 8114182679,
+    "ultra_style/input_tokens": 8114608556,
+    "openmath/blocks": 3420000,
+    "openmath/targets": 5284800824,
+    "openmath/input_tokens": 5286561887,
+    "rewrite6/blocks": 1710000,
+    "rewrite6/targets": 2803276708,
+    "rewrite6/input_tokens": 2806350227
   },
   "sources": {
     "climbmix": {
-      "page": 1849,
-      "row": 879,
+      "page": 1873,
+      "row": 3681,
       "start": {
-        "file": 89,
+        "file": 90,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 41,
+          "row_group": 54,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 7574383
+      "documents": 7675489
     },
     "rewrite6": {
-      "page": 1,
-      "row": 3226,
+      "page": 15,
+      "row": 3703,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 4096
+          "row": 61440
         }
       },
       "epoch": 21,
       "offset": 0,
-      "documents": 4320806
+      "documents": 4378627
     },
     "openmath": {
-      "page": 1155,
-      "row": 2141,
+      "page": 1170,
+      "row": 3787,
       "start": {
         "file": 10,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 364,
-          "row": 380
+          "row_group": 425,
+          "row": 820
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 4733021
+      "documents": 4796107
     },
     "ultra_qa": {
-      "page": 499,
-      "row": 2492,
+      "page": 506,
+      "row": 1064,
       "start": {
         "file": 3,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 32909
+          "row": 61581
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 2046396
+      "documents": 2073640
     },
     "ultra_style": {
-      "page": 762,
-      "row": 3326,
+      "page": 773,
+      "row": 137,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 4,
-          "row": 40896
+          "row_group": 5,
+          "row": 95
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3124478
+      "documents": 3166345
     },
     "cortex": {
-      "page": 98190,
-      "row": 71,
+      "page": 99499,
+      "row": 85,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 12568391
+      "documents": 12735957
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 375000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "6610966b863295dd906696faac88a04c6ff0496a879a152f0273bb1b9d396255", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "406bc235a118a90d6ce9168e0c61f5aeee892c2cd7892ea2dfe941963fb47bc3", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "e3923fa86352f6d878fd8d37b6b9fc374577f52be522203b0b865badf630da15", "shard_scheduler.json": "3446bd4676a0432db7c222c8eed67dd16a89c65bbd51ec69a9afb2a131cff95e", "coverage.json": "eb5f67654bace9fef82be9a9d8a490ab08cd158a627ef4268f68ce6a57faf403", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000375000"}
+{"step": 380000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "c1d74f83e24ba5ed4a98e7a8650879fde80d5e46dcea28dfeb99c4ff2c71c650", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "96586ee6a38bb3d010196afc962d100dcda3fae0e33facf3c5450cee743fa5c6", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "e3923fa86352f6d878fd8d37b6b9fc374577f52be522203b0b865badf630da15", "shard_scheduler.json": "99b8b50757cc145af0cc487dcd657bdacdcccce20a0ed8fcd882dd2d990666a8", "coverage.json": "1bcb49c8690d4fac9adfe028153b744eff8144e6c17080d648fa5eaf6ebf1532", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000380000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6610966b863295dd906696faac88a04c6ff0496a879a152f0273bb1b9d396255
+oid sha256:c1d74f83e24ba5ed4a98e7a8650879fde80d5e46dcea28dfeb99c4ff2c71c650
 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": 75451.54704015724,
-      "required_pages_per_second_upper_estimate": 0.006539948867243167
+      "required_token_positions_per_second": 75899.45532405979,
+      "required_pages_per_second_upper_estimate": 0.006578772422078673
     },
     {
       "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": 9431.443380019655,
-      "required_pages_per_second_upper_estimate": 0.0037623437769346
+      "required_token_positions_per_second": 9487.431915507474,
+      "required_pages_per_second_upper_estimate": 0.003784678440843894
     },
     {
       "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": 18862.88676003931,
-      "required_pages_per_second_upper_estimate": 0.004363596876466488
+      "required_token_positions_per_second": 18974.863831014947,
+      "required_pages_per_second_upper_estimate": 0.004389500801107561
     },
     {
       "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": 28294.330140058966,
-      "required_pages_per_second_upper_estimate": 0.00172297625665208
+      "required_token_positions_per_second": 28462.295746522417,
+      "required_pages_per_second_upper_estimate": 0.0017332044808382734
     },
     {
       "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": 28294.330140058966,
-      "required_pages_per_second_upper_estimate": 0.002178175857900508
+      "required_token_positions_per_second": 28462.295746522417,
+      "required_pages_per_second_upper_estimate": 0.0021911063152446216
     },
     {
       "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": 28294.330140058966,
-      "required_pages_per_second_upper_estimate": 0.3513208853081064
+      "required_token_positions_per_second": 28462.295746522417,
+      "required_pages_per_second_upper_estimate": 0.35340645599565934
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,409 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 717798,
-    "producer_batch_seconds": 0.11261634734350423,
-    "producer_p95_seconds": 0.27807193199987523,
-    "gpu_update_seconds": 0.9771569025716607,
-    "producer_batches_per_second": 8.879705509802974,
-    "gpu_batches_per_second": 1.023377102866716,
-    "producer_headroom": 8.676865531707582,
+    "producer_batch_seconds": 0.08713245738275077,
+    "producer_p95_seconds": 0.28750059899903135,
+    "gpu_update_seconds": 0.9713903701312671,
+    "producer_batches_per_second": 11.47677949225343,
+    "gpu_batches_per_second": 1.0294522477764185,
+    "producer_headroom": 11.148433078894994,
     "sustainable": true,
-    "buffer_seconds": 60.58372795944296,
+    "buffer_seconds": 60.22620294813856,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 83,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 413636.13457235816,
-    "upload_bytes_per_second": 5862199.890527139,
+    "generation_bytes_per_second": 248590.9986309057,
+    "upload_bytes_per_second": 5086383.092773639,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 88.75668552282389,
+    "estimated_full_shards_per_hour": 53.34184140391711,
     "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": 2.0509548539994285,
-      "bytes": 12950750,
-      "pages": 26,
-      "shards": 4
+      "seconds": 1.0379702050013293,
+      "bytes": 6386142,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.8878694479990372,
-      "bytes": 1157611,
+      "seconds": 1.1697385600018606,
+      "bytes": 11844982,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.079595678998885,
+      "bytes": 1118845,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.0454248690002714,
-      "bytes": 6130575,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.1773780209987308,
+      "bytes": 12342839,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.3716071430008014,
-      "bytes": 11837179,
+      "seconds": 1.146244535997539,
+      "bytes": 6228835,
       "pages": 36,
-      "shards": 4
+      "shards": 2
     },
     {
-      "seconds": 0.8982555259990477,
-      "bytes": 1117404,
-      "pages": 34,
+      "seconds": 1.0997196959979192,
+      "bytes": 878203,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 1.2563196339997376,
-      "bytes": 7398482,
-      "pages": 29,
+      "seconds": 1.3262504000012996,
+      "bytes": 7612878,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.2963822889996663,
-      "bytes": 6635646,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2522562880003534,
-      "bytes": 9539459,
-      "pages": 41,
+      "seconds": 1.397075862998463,
+      "bytes": 11930697,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 0.8906719839997095,
-      "bytes": 2274847,
-      "pages": 27,
-      "shards": 2
-    },
-    {
-      "seconds": 1.014783542999794,
-      "bytes": 6341777,
+      "seconds": 1.03103892800209,
+      "bytes": 6422195,
       "pages": 35,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.1726115760011453,
-      "bytes": 10149882,
+      "seconds": 1.064075487000082,
+      "bytes": 6211136,
       "pages": 35,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.0351456819989835,
-      "bytes": 2694003,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.7981215300023905,
+      "bytes": 910382,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.6696938850000151,
-      "bytes": 11635327,
-      "pages": 30,
+      "seconds": 1.0455480339987844,
+      "bytes": 7415331,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0197001609994913,
-      "bytes": 7722133,
-      "pages": 38,
+      "seconds": 1.2304811680005514,
+      "bytes": 10294757,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0303949059998558,
-      "bytes": 5043636,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.5211265639991325,
+      "bytes": 8178789,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.2701704230003088,
-      "bytes": 6145157,
-      "pages": 35,
+      "seconds": 0.9424325320032949,
+      "bytes": 6190430,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1758933869987231,
-      "bytes": 2380578,
-      "pages": 31,
+      "seconds": 0.982011802996567,
+      "bytes": 5926739,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.1564406300003611,
-      "bytes": 5972956,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.048674845998903,
+      "bytes": 7634002,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.3616979350008478,
-      "bytes": 18065144,
-      "pages": 42,
-      "shards": 5
+      "seconds": 1.2676819320004142,
+      "bytes": 6003365,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 0.8625212450006075,
-      "bytes": 879863,
-      "pages": 25,
+      "seconds": 0.7857203480016324,
+      "bytes": 1089066,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.04706289600108,
-      "bytes": 6208354,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.0898243610026839,
+      "bytes": 13191576,
+      "pages": 31,
+      "shards": 4
     },
     {
-      "seconds": 0.9189691079991462,
-      "bytes": 2568805,
-      "pages": 36,
+      "seconds": 0.8497731899988139,
+      "bytes": 5387577,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.252516081000067,
-      "bytes": 9866864,
-      "pages": 30,
+      "seconds": 1.2923507409977901,
+      "bytes": 7469632,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0568891879993316,
-      "bytes": 6084196,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 0.8473804010009189,
-      "bytes": 2318684,
-      "pages": 28,
+      "seconds": 1.187904915001127,
+      "bytes": 6100274,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1035010350005905,
-      "bytes": 11692213,
-      "pages": 31,
-      "shards": 3
+      "seconds": 0.7990551539987791,
+      "bytes": 562082,
+      "pages": 17,
+      "shards": 1
     },
     {
-      "seconds": 1.0510043949998362,
-      "bytes": 4790558,
-      "pages": 36,
+      "seconds": 0.8617004240004462,
+      "bytes": 4032617,
+      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 1.1652991789997031,
-      "bytes": 7620124,
-      "pages": 34,
+      "seconds": 1.6352041230020404,
+      "bytes": 7545726,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0473349340009008,
-      "bytes": 5879474,
-      "pages": 30,
+      "seconds": 1.4504600340005709,
+      "bytes": 11937829,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.929622351000944,
-      "bytes": 6601933,
-      "pages": 39,
+      "seconds": 1.2463495720003266,
+      "bytes": 2130021,
+      "pages": 22,
+      "shards": 2
+    },
+    {
+      "seconds": 2.160228343000199,
+      "bytes": 11006785,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 2.1250439309987996,
-      "bytes": 6028843,
-      "pages": 31,
+      "seconds": 1.0801306080029462,
+      "bytes": 6178099,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9982244909988367,
-      "bytes": 6934101,
-      "pages": 35,
+      "seconds": 1.0223706500000844,
+      "bytes": 2585543,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.4433788420028577,
-      "bytes": 7611195,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.1590529789973516,
+      "bytes": 5922158,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.260750511999504,
-      "bytes": 10895855,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.1085641130011936,
+      "bytes": 12447049,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 1.124983618999977,
-      "bytes": 2583658,
-      "pages": 36,
+      "seconds": 1.0412157949976972,
+      "bytes": 5733012,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.0313133240015304,
-      "bytes": 5497040,
-      "pages": 26,
+      "seconds": 1.004709222001111,
+      "bytes": 6095634,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0232497649994912,
-      "bytes": 1108747,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.292045906000567,
-      "bytes": 13335425,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.0117634249982075,
+      "bytes": 2615219,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.1053190409984381,
-      "bytes": 10451885,
-      "pages": 40,
+      "seconds": 1.2512058009997418,
+      "bytes": 10378380,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.8453881149980589,
-      "bytes": 832969,
-      "pages": 26,
+      "seconds": 2.6139940620014386,
+      "bytes": 1187973,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.4130192459997488,
-      "bytes": 7383726,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.1084424000000581,
+      "bytes": 13383772,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.007008101001702,
-      "bytes": 6024431,
+      "seconds": 1.2797600039993995,
+      "bytes": 6001788,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.0379702050013293,
-      "bytes": 6386142,
-      "pages": 34,
+      "seconds": 1.0512617659987882,
+      "bytes": 6612201,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.1697385600018606,
-      "bytes": 11844982,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.0170475559971237,
+      "bytes": 6061997,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.079595678998885,
-      "bytes": 1118845,
-      "pages": 34,
+      "seconds": 0.9019806840005913,
+      "bytes": 1136025,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.1773780209987308,
-      "bytes": 12342839,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.5857687079987954,
+      "bytes": 7492427,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.146244535997539,
-      "bytes": 6228835,
-      "pages": 36,
+      "seconds": 1.18117357400115,
+      "bytes": 5405796,
+      "pages": 19,
       "shards": 2
     },
     {
-      "seconds": 1.0997196959979192,
-      "bytes": 878203,
-      "pages": 25,
-      "shards": 1
+      "seconds": 2.5150336740007333,
+      "bytes": 12238900,
+      "pages": 44,
+      "shards": 3
     },
     {
-      "seconds": 1.3262504000012996,
-      "bytes": 7612878,
+      "seconds": 2.105870221999794,
+      "bytes": 7597429,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.397075862998463,
-      "bytes": 11930697,
+      "seconds": 4.179794984000182,
+      "bytes": 6264644,
       "pages": 36,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.03103892800209,
-      "bytes": 6422195,
-      "pages": 35,
-      "shards": 3
+      "seconds": 0.9406074870021257,
+      "bytes": 762678,
+      "pages": 21,
+      "shards": 1
     },
     {
-      "seconds": 1.064075487000082,
-      "bytes": 6211136,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.7635130470007425,
+      "bytes": 11843733,
+      "pages": 46,
+      "shards": 5
     },
     {
-      "seconds": 0.7981215300023905,
-      "bytes": 910382,
-      "pages": 26,
+      "seconds": 0.9203152180016332,
+      "bytes": 582089,
+      "pages": 18,
       "shards": 1
     },
     {
-      "seconds": 1.0455480339987844,
-      "bytes": 7415331,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.208456311996997,
+      "bytes": 13456782,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.2304811680005514,
-      "bytes": 10294757,
-      "pages": 37,
-      "shards": 3
+      "seconds": 0.7929018860013457,
+      "bytes": 738445,
+      "pages": 21,
+      "shards": 1
     },
     {
-      "seconds": 1.5211265639991325,
-      "bytes": 8178789,
-      "pages": 36,
+      "seconds": 1.138339361998078,
+      "bytes": 10270327,
+      "pages": 46,
       "shards": 3
     },
     {
-      "seconds": 0.9424325320032949,
-      "bytes": 6190430,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.5007521030020143,
+      "bytes": 7605566,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 0.982011802996567,
-      "bytes": 5926739,
-      "pages": 33,
-      "shards": 2
+      "seconds": 0.8186183350007923,
+      "bytes": 971258,
+      "pages": 28,
+      "shards": 1
     },
     {
-      "seconds": 1.048674845998903,
-      "bytes": 7634002,
-      "pages": 29,
+      "seconds": 1.5642075370014936,
+      "bytes": 10184682,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.2676819320004142,
-      "bytes": 6003365,
-      "pages": 36,
+      "seconds": 1.1651681650000683,
+      "bytes": 8287753,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.7857203480016324,
-      "bytes": 1089066,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.1439757010011817,
+      "bytes": 6251248,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.0898243610026839,
-      "bytes": 13191576,
-      "pages": 31,
-      "shards": 4
+      "seconds": 0.9833170429992606,
+      "bytes": 2688996,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.8497731899988139,
-      "bytes": 5387577,
-      "pages": 34,
+      "seconds": 1.4986307230028615,
+      "bytes": 10640144,
+      "pages": 35,
+      "shards": 3
+    },
+    {
+      "seconds": 3.6133480419994157,
+      "bytes": 5871337,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.2923507409977901,
-      "bytes": 7469632,
+      "seconds": 1.06491817099959,
+      "bytes": 2604937,
       "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.2623599489998014,
+      "bytes": 11727417,
+      "pages": 35,
       "shards": 3
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:406bc235a118a90d6ce9168e0c61f5aeee892c2cd7892ea2dfe941963fb47bc3
-size 721764351
+oid sha256:96586ee6a38bb3d010196afc962d100dcda3fae0e33facf3c5450cee743fa5c6
+size 721779135