Qwen90M step 375000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 370000,
+  "step": 375000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "27d38bcb78795a7f65759cb56e8869b8f3f0b60c43b03779747aade9b5f5dfb1",
+    "model.safetensors": "6610966b863295dd906696faac88a04c6ff0496a879a152f0273bb1b9d396255",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "bb0f4782fb80c8565ac6be62823151fb471a6fd6261d4d56ac9296ed0dcad246",
+    "trainer.pt": "406bc235a118a90d6ce9168e0c61f5aeee892c2cd7892ea2dfe941963fb47bc3",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "e3923fa86352f6d878fd8d37b6b9fc374577f52be522203b0b865badf630da15",
-    "shard_scheduler.json": "b624032da86b57ae0825c33e90bb460d2485d24bcaf10e56c24f0be2d7391d4f",
-    "coverage.json": "6910c159e150a2a8f8c0838a7ccd1c7e5752fb223c3f3f4a879ec42144d06cf8",
+    "shard_scheduler.json": "3446bd4676a0432db7c222c8eed67dd16a89c65bbd51ec69a9afb2a131cff95e",
+    "coverage.json": "eb5f67654bace9fef82be9a9d8a490ab08cd158a627ef4268f68ce6a57faf403",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 370000,
+  "step": 375000,
   "stats": {
-    "climbmix/blocks": 13320000,
-    "climbmix/targets": 22265879849,
-    "climbmix/input_tokens": 22267290588,
-    "ultra_qa/blocks": 4995000,
-    "ultra_qa/targets": 8148669403,
-    "ultra_qa/input_tokens": 8148685149,
-    "cortex/blocks": 4995000,
-    "cortex/targets": 190015600,
-    "cortex/input_tokens": 7940414338,
-    "ultra_style/blocks": 4995000,
-    "ultra_style/targets": 7900652448,
-    "ultra_style/input_tokens": 7901067026,
-    "openmath/blocks": 3330000,
-    "openmath/targets": 5145705012,
-    "openmath/input_tokens": 5147419580,
-    "rewrite6/blocks": 1665000,
-    "rewrite6/targets": 2729490219,
-    "rewrite6/input_tokens": 2732482833
+    "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
   },
   "sources": {
     "climbmix": {
-      "page": 1824,
-      "row": 1985,
+      "page": 1849,
+      "row": 879,
       "start": {
-        "file": 88,
+        "file": 89,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 24,
+          "row_group": 41,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 7473089
+      "documents": 7574383
     },
     "rewrite6": {
-      "page": 37,
-      "row": 3836,
+      "page": 1,
+      "row": 3226,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 151552
+          "row": 4096
         }
       },
-      "epoch": 20,
+      "epoch": 21,
       "offset": 0,
-      "documents": 4263468
+      "documents": 4320806
     },
     "openmath": {
-      "page": 1140,
-      "row": 244,
+      "page": 1155,
+      "row": 2141,
       "start": {
         "file": 10,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 302,
-          "row": 940
+          "row_group": 364,
+          "row": 380
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 4669684
+      "documents": 4733021
     },
     "ultra_qa": {
-      "page": 492,
-      "row": 3817,
+      "page": 499,
+      "row": 2492,
       "start": {
         "file": 3,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 4237
+          "row": 32909
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 2019049
+      "documents": 2046396
     },
     "ultra_style": {
-      "page": 752,
-      "row": 2480,
+      "page": 762,
+      "row": 3326,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 85359
+          "row_group": 4,
+          "row": 40896
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3082672
+      "documents": 3124478
     },
     "cortex": {
-      "page": 96880,
-      "row": 25,
+      "page": 98190,
+      "row": 71,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 12400665
+      "documents": 12568391
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 370000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "27d38bcb78795a7f65759cb56e8869b8f3f0b60c43b03779747aade9b5f5dfb1", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "bb0f4782fb80c8565ac6be62823151fb471a6fd6261d4d56ac9296ed0dcad246", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "e3923fa86352f6d878fd8d37b6b9fc374577f52be522203b0b865badf630da15", "shard_scheduler.json": "b624032da86b57ae0825c33e90bb460d2485d24bcaf10e56c24f0be2d7391d4f", "coverage.json": "6910c159e150a2a8f8c0838a7ccd1c7e5752fb223c3f3f4a879ec42144d06cf8", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000370000"}
+{"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"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:27d38bcb78795a7f65759cb56e8869b8f3f0b60c43b03779747aade9b5f5dfb1
+oid sha256:6610966b863295dd906696faac88a04c6ff0496a879a152f0273bb1b9d396255
 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": 75577.40725271897,
-      "required_pages_per_second_upper_estimate": 0.0065508581114782265
+      "required_token_positions_per_second": 75451.54704015724,
+      "required_pages_per_second_upper_estimate": 0.006539948867243167
     },
     {
       "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": 9447.175906589871,
-      "required_pages_per_second_upper_estimate": 0.003768619717005693
+      "required_token_positions_per_second": 9431.443380019655,
+      "required_pages_per_second_upper_estimate": 0.0037623437769346
     },
     {
       "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": 18894.351813179743,
-      "required_pages_per_second_upper_estimate": 0.004370875762744505
+      "required_token_positions_per_second": 18862.88676003931,
+      "required_pages_per_second_upper_estimate": 0.004363596876466488
     },
     {
       "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": 28341.527719769612,
-      "required_pages_per_second_upper_estimate": 0.001725850341629895
+      "required_token_positions_per_second": 28294.330140058966,
+      "required_pages_per_second_upper_estimate": 0.00172297625665208
     },
     {
       "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": 28341.527719769612,
-      "required_pages_per_second_upper_estimate": 0.002181809258237896
+      "required_token_positions_per_second": 28294.330140058966,
+      "required_pages_per_second_upper_estimate": 0.002178175857900508
     },
     {
       "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": 28341.527719769612,
-      "required_pages_per_second_upper_estimate": 0.35190692128797463
+      "required_token_positions_per_second": 28294.330140058966,
+      "required_pages_per_second_upper_estimate": 0.3513208853081064
     }
   ],
   "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": 717798,
-    "producer_batch_seconds": 0.08124377494536361,
-    "producer_p95_seconds": 0.21642768499987142,
-    "gpu_update_seconds": 0.9755296282322197,
-    "producer_batches_per_second": 12.30863534680041,
-    "gpu_batches_per_second": 1.0250841912532414,
-    "producer_headroom": 12.00743846391016,
+    "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,
     "sustainable": true,
-    "buffer_seconds": 60.48283695039762,
+    "buffer_seconds": 60.58372795944296,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 80,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 312252.486873165,
-    "upload_bytes_per_second": 5783967.833540881,
+    "generation_bytes_per_second": 413636.13457235816,
+    "upload_bytes_per_second": 5862199.890527139,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 67.00211481710637,
+    "estimated_full_shards_per_hour": 88.75668552282389,
     "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": 0.9784717390002697,
-      "bytes": 7343882,
-      "pages": 29,
-      "shards": 3
+      "seconds": 2.0509548539994285,
+      "bytes": 12950750,
+      "pages": 26,
+      "shards": 4
     },
     {
-      "seconds": 0.8120560110000952,
-      "bytes": 1154123,
-      "pages": 33,
+      "seconds": 0.8878694479990372,
+      "bytes": 1157611,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.3542424540000866,
-      "bytes": 9851710,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.0454248690002714,
+      "bytes": 6130575,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.397486900000331,
-      "bytes": 13488820,
-      "pages": 39,
+      "seconds": 1.3716071430008014,
+      "bytes": 11837179,
+      "pages": 36,
       "shards": 4
     },
     {
-      "seconds": 0.9587566239988519,
-      "bytes": 1052447,
-      "pages": 33,
+      "seconds": 0.8982555259990477,
+      "bytes": 1117404,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.538976370000455,
-      "bytes": 12078751,
-      "pages": 37,
-      "shards": 5
-    },
-    {
-      "seconds": 0.9161946559997887,
-      "bytes": 856768,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.2563196339997376,
+      "bytes": 7398482,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.2508889310011,
-      "bytes": 6008939,
-      "pages": 36,
+      "seconds": 1.2963822889996663,
+      "bytes": 6635646,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.107625275000828,
-      "bytes": 7585534,
-      "pages": 37,
+      "seconds": 1.2522562880003534,
+      "bytes": 9539459,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.0756502530002763,
-      "bytes": 6680691,
-      "pages": 29,
+      "seconds": 0.8906719839997095,
+      "bytes": 2274847,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.0957601480004087,
-      "bytes": 10587203,
-      "pages": 33,
+      "seconds": 1.014783542999794,
+      "bytes": 6341777,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1726115760011453,
+      "bytes": 10149882,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.9222303849983291,
-      "bytes": 2581513,
+      "seconds": 1.0351456819989835,
+      "bytes": 2694003,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2870966139998927,
-      "bytes": 6072045,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.6696938850000151,
+      "bytes": 11635327,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.1727791039993463,
-      "bytes": 7506500,
-      "pages": 37,
+      "seconds": 1.0197001609994913,
+      "bytes": 7722133,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0041788279995671,
-      "bytes": 4770921,
-      "pages": 34,
+      "seconds": 1.0303949059998558,
+      "bytes": 5043636,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.17236374300046,
-      "bytes": 5601597,
-      "pages": 19,
+      "seconds": 1.2701704230003088,
+      "bytes": 6145157,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1996037619992421,
-      "bytes": 8790333,
-      "pages": 45,
-      "shards": 3
+      "seconds": 1.1758933869987231,
+      "bytes": 2380578,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 0.831416236998848,
-      "bytes": 5844217,
-      "pages": 30,
+      "seconds": 1.1564406300003611,
+      "bytes": 5972956,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.3119218800002272,
-      "bytes": 12680620,
-      "pages": 37,
-      "shards": 4
+      "seconds": 1.3616979350008478,
+      "bytes": 18065144,
+      "pages": 42,
+      "shards": 5
     },
     {
-      "seconds": 0.8689146019987675,
-      "bytes": 1090693,
-      "pages": 33,
+      "seconds": 0.8625212450006075,
+      "bytes": 879863,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 0.9899479589985276,
-      "bytes": 6233203,
+      "seconds": 1.04706289600108,
+      "bytes": 6208354,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8259697120010969,
-      "bytes": 2261871,
-      "pages": 24,
+      "seconds": 0.9189691079991462,
+      "bytes": 2568805,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 2.0209359849995963,
-      "bytes": 15406607,
-      "pages": 37,
-      "shards": 4
+      "seconds": 1.252516081000067,
+      "bytes": 9866864,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.1222496679984033,
-      "bytes": 5615350,
-      "pages": 27,
+      "seconds": 1.0568891879993316,
+      "bytes": 6084196,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.8820894919990678,
-      "bytes": 2581402,
-      "pages": 34,
+      "seconds": 0.8473804010009189,
+      "bytes": 2318684,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0837803860013082,
-      "bytes": 6217801,
+      "seconds": 1.1035010350005905,
+      "bytes": 11692213,
+      "pages": 31,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0510043949998362,
+      "bytes": 4790558,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.5358513270002732,
-      "bytes": 5360685,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1652991789997031,
+      "bytes": 7620124,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.0740509600000223,
-      "bytes": 7543060,
-      "pages": 37,
+      "seconds": 1.0473349340009008,
+      "bytes": 5879474,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.1867137099998217,
-      "bytes": 11783484,
-      "pages": 32,
+      "seconds": 0.929622351000944,
+      "bytes": 6601933,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.1010340770008042,
-      "bytes": 2510601,
+      "seconds": 2.1250439309987996,
+      "bytes": 6028843,
+      "pages": 31,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9982244909988367,
+      "bytes": 6934101,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.0974370109997835,
-      "bytes": 10722524,
-      "pages": 40,
+      "seconds": 1.4433788420028577,
+      "bytes": 7611195,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.196992418001173,
-      "bytes": 774765,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.260750511999504,
+      "bytes": 10895855,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.1212218569999095,
-      "bytes": 7621728,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.124983618999977,
+      "bytes": 2583658,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.0421719010009838,
-      "bytes": 6283978,
-      "pages": 35,
+      "seconds": 1.0313133240015304,
+      "bytes": 5497040,
+      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 1.2377747039990936,
-      "bytes": 12765065,
-      "pages": 37,
+      "seconds": 1.0232497649994912,
+      "bytes": 1108747,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.292045906000567,
+      "bytes": 13335425,
+      "pages": 35,
       "shards": 4
     },
     {
-      "seconds": 1.0026179279993812,
-      "bytes": 5878114,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1053190409984381,
+      "bytes": 10451885,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 0.8248095879989705,
-      "bytes": 1051965,
-      "pages": 33,
+      "seconds": 0.8453881149980589,
+      "bytes": 832969,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.2737176890004775,
-      "bytes": 7770049,
-      "pages": 38,
+      "seconds": 1.4130192459997488,
+      "bytes": 7383726,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1648699769993982,
-      "bytes": 6106287,
-      "pages": 25,
+      "seconds": 1.007008101001702,
+      "bytes": 6024431,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.0237944010004867,
-      "bytes": 5653726,
-      "pages": 43,
-      "shards": 2
+      "seconds": 1.0379702050013293,
+      "bytes": 6386142,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 2.0509548539994285,
-      "bytes": 12950750,
-      "pages": 26,
-      "shards": 4
+      "seconds": 1.1697385600018606,
+      "bytes": 11844982,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.8878694479990372,
-      "bytes": 1157611,
+      "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
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bb0f4782fb80c8565ac6be62823151fb471a6fd6261d4d56ac9296ed0dcad246
-size 721784447
+oid sha256:406bc235a118a90d6ce9168e0c61f5aeee892c2cd7892ea2dfe941963fb47bc3
+size 721764351