Qwen90M step 345000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 340000,
+  "step": 345000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "cfe85d751372f6d04e4c83673f7f4a64d053832558f082b05e3f10cddd637ea0",
+    "model.safetensors": "1280a8081ba139ed35dc86617f2d00c744f61dc8d28e7148fdcf6e8ecb03db3e",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "dee880db6477edfe4f23e2084b87d73ef0b8e70abdddbdfbbee6e8595a7fc54b",
+    "trainer.pt": "b48df57d11cb4744bb19a078c1bcb987e665797abd0171d45df1688e70bd7bcf",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "90ad39d684ac792f235cbfbeef498d4c81f09977d8f89fb8c3bdd41e515f347a",
-    "shard_scheduler.json": "9679ee0365d0b904eb8a24d5521d34bfd003fd3e82dd158575c1b3b398fa32a1",
-    "coverage.json": "5106d799b368c806803fb857e1c440928c795ad2f28b3557233a0ae0605df710",
+    "shard_scheduler.json": "ed5ff406bb283085db0d0d6ae907c0411de8aeb884c3d97dca7e818c32facac2",
+    "coverage.json": "2b001d842650d60a89eccf53c95a56122f777cfa328d77fd27f8403912f812c5",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 340000,
+  "step": 345000,
   "stats": {
-    "climbmix/blocks": 12240000,
-    "climbmix/targets": 20460879849,
-    "climbmix/input_tokens": 20462176730,
-    "ultra_qa/blocks": 4590000,
-    "ultra_qa/targets": 7488182065,
-    "ultra_qa/input_tokens": 7488196546,
-    "cortex/blocks": 4590000,
-    "cortex/targets": 174627193,
-    "cortex/input_tokens": 7296594720,
-    "ultra_style/blocks": 4590000,
-    "ultra_style/targets": 7260228205,
-    "ultra_style/input_tokens": 7260609491,
-    "openmath/blocks": 3060000,
-    "openmath/targets": 4728475624,
-    "openmath/input_tokens": 4730051742,
-    "rewrite6/blocks": 1530000,
-    "rewrite6/targets": 2508218738,
-    "rewrite6/input_tokens": 2510968592
+    "climbmix/blocks": 12420000,
+    "climbmix/targets": 20761699264,
+    "climbmix/input_tokens": 20763015181,
+    "ultra_qa/blocks": 4657500,
+    "ultra_qa/targets": 7598219171,
+    "ultra_qa/input_tokens": 7598233863,
+    "cortex/blocks": 4657500,
+    "cortex/targets": 177190627,
+    "cortex/input_tokens": 7403880935,
+    "ultra_style/blocks": 4657500,
+    "ultra_style/targets": 7366972233,
+    "ultra_style/input_tokens": 7367358771,
+    "openmath/blocks": 3105000,
+    "openmath/targets": 4798065277,
+    "openmath/input_tokens": 4799664519,
+    "rewrite6/blocks": 1552500,
+    "rewrite6/targets": 2545077728,
+    "rewrite6/input_tokens": 2547868299
   },
   "sources": {
     "climbmix": {
-      "page": 1676,
-      "row": 2328,
+      "page": 1701,
+      "row": 907,
       "start": {
-        "file": 81,
+        "file": 82,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 10,
+          "row_group": 27,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 6867224
+      "documents": 6968203
     },
     "rewrite6": {
-      "page": 3,
-      "row": 2617,
+      "page": 17,
+      "row": 3186,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 12288
+          "row": 69632
         }
       },
       "epoch": 19,
-      "offset": 0,
-      "documents": 3917581
+      "offset": 2048,
+      "documents": 3975494
     },
     "openmath": {
-      "page": 1047,
-      "row": 3249,
+      "page": 1063,
+      "row": 868,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 358,
-          "row": 662
+          "row_group": 424,
+          "row": 198
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 4291761
+      "offset": 0,
+      "documents": 4354916
     },
     "ultra_qa": {
-      "page": 452,
-      "row": 3966,
+      "page": 459,
+      "row": 2604,
       "start": {
         "file": 3,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 34139
+          "row": 62811
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 1855358
+      "offset": 4096,
+      "documents": 1882668
     },
     "ultra_style": {
-      "page": 691,
-      "row": 2767,
+      "page": 701,
+      "row": 2909,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 5110
+          "row": 46070
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 2833103
+      "documents": 2874205
     },
     "cortex": {
-      "page": 89025,
-      "row": 35,
+      "page": 90333,
+      "row": 88,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 11395235
+      "documents": 11562712
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 340000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "cfe85d751372f6d04e4c83673f7f4a64d053832558f082b05e3f10cddd637ea0", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "dee880db6477edfe4f23e2084b87d73ef0b8e70abdddbdfbbee6e8595a7fc54b", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "90ad39d684ac792f235cbfbeef498d4c81f09977d8f89fb8c3bdd41e515f347a", "shard_scheduler.json": "9679ee0365d0b904eb8a24d5521d34bfd003fd3e82dd158575c1b3b398fa32a1", "coverage.json": "5106d799b368c806803fb857e1c440928c795ad2f28b3557233a0ae0605df710", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000340000"}
+{"step": 345000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "1280a8081ba139ed35dc86617f2d00c744f61dc8d28e7148fdcf6e8ecb03db3e", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "b48df57d11cb4744bb19a078c1bcb987e665797abd0171d45df1688e70bd7bcf", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "90ad39d684ac792f235cbfbeef498d4c81f09977d8f89fb8c3bdd41e515f347a", "shard_scheduler.json": "ed5ff406bb283085db0d0d6ae907c0411de8aeb884c3d97dca7e818c32facac2", "coverage.json": "2b001d842650d60a89eccf53c95a56122f777cfa328d77fd27f8403912f812c5", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000345000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cfe85d751372f6d04e4c83673f7f4a64d053832558f082b05e3f10cddd637ea0
+oid sha256:1280a8081ba139ed35dc86617f2d00c744f61dc8d28e7148fdcf6e8ecb03db3e
 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": 75407.3266986464,
-      "required_pages_per_second_upper_estimate": 0.006536115960116439
+      "required_token_positions_per_second": 75294.19321551842,
+      "required_pages_per_second_upper_estimate": 0.006526309836533097
     },
     {
       "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": 9425.9158373308,
-      "required_pages_per_second_upper_estimate": 0.0037601387575118875
+      "required_token_positions_per_second": 9411.774151939802,
+      "required_pages_per_second_upper_estimate": 0.00375449742777238
     },
     {
       "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": 18851.8316746616,
-      "required_pages_per_second_upper_estimate": 0.004361039477027212
+      "required_token_positions_per_second": 18823.548303879605,
+      "required_pages_per_second_upper_estimate": 0.004354496616967123
     },
     {
       "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": 28277.7475119924,
-      "required_pages_per_second_upper_estimate": 0.0017219664616051516
+      "required_token_positions_per_second": 28235.32245581941,
+      "required_pages_per_second_upper_estimate": 0.0017193829982712838
     },
     {
       "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": 28277.7475119924,
-      "required_pages_per_second_upper_estimate": 0.002176899281288291
+      "required_token_positions_per_second": 28235.32245581941,
+      "required_pages_per_second_upper_estimate": 0.0021736332830240225
     },
     {
       "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": 28277.7475119924,
-      "required_pages_per_second_upper_estimate": 0.35111498456600565
+      "required_token_positions_per_second": 28235.32245581941,
+      "required_pages_per_second_upper_estimate": 0.3505882073558664
     }
   ],
   "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": 690670,
-    "producer_batch_seconds": 0.0937007041719653,
-    "producer_p95_seconds": 0.30818277499929536,
-    "gpu_update_seconds": 0.9777299266242714,
-    "producer_batches_per_second": 10.672278387201215,
-    "gpu_batches_per_second": 1.022777326099262,
-    "producer_headroom": 10.434605964432041,
+    "producer_batch_seconds": 0.11289917382015346,
+    "producer_p95_seconds": 0.3555639389996941,
+    "gpu_update_seconds": 0.9791990172331693,
+    "producer_batches_per_second": 8.85746074274187,
+    "gpu_batches_per_second": 1.0212428550281905,
+    "producer_headroom": 8.673216854474216,
     "sustainable": true,
-    "buffer_seconds": 60.61925545070483,
+    "buffer_seconds": 60.71033906845649,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 83,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 364474.32041737513,
-    "upload_bytes_per_second": 5260233.793448558,
+    "generation_bytes_per_second": 312717.6648123108,
+    "upload_bytes_per_second": 4152064.367452919,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 78.20770463362636,
+    "estimated_full_shards_per_hour": 67.10193117405885,
     "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": 1.2185796390003816,
-      "bytes": 6487750,
-      "pages": 35,
+      "seconds": 1.2690833449996717,
+      "bytes": 2379532,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.3924951020017033,
-      "bytes": 2562340,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.429218111999944,
-      "bytes": 10705946,
-      "pages": 39,
+      "seconds": 1.1314427079996676,
+      "bytes": 10762955,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.0223479320011393,
-      "bytes": 6489223,
-      "pages": 24,
+      "seconds": 1.0500791910017142,
+      "bytes": 5691792,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.180141814002127,
-      "bytes": 7534736,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.035739069000556,
+      "bytes": 2751840,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.0304640750000544,
-      "bytes": 6585841,
-      "pages": 35,
+      "seconds": 1.1663796350003395,
+      "bytes": 6310887,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9793735439998272,
-      "bytes": 7348514,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.171979762999399,
+      "bytes": 12042987,
+      "pages": 27,
+      "shards": 4
     },
     {
-      "seconds": 1.1231975430018792,
-      "bytes": 5800198,
-      "pages": 27,
+      "seconds": 1.1372878729998774,
+      "bytes": 6231473,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.8728474010022182,
-      "bytes": 1156879,
-      "pages": 34,
-      "shards": 1
+      "seconds": 0.9854615350013773,
+      "bytes": 5689074,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.1430372309987433,
-      "bytes": 13489384,
-      "pages": 40,
-      "shards": 4
+      "seconds": 0.9606277580023743,
+      "bytes": 2676212,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.410483870000462,
-      "bytes": 10063648,
-      "pages": 28,
+      "seconds": 1.570391033001215,
+      "bytes": 9821415,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0145048790000146,
-      "bytes": 1134517,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.1360236319997057,
+      "bytes": 2731200,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 3.9685035570000764,
-      "bytes": 7473998,
-      "pages": 38,
+      "seconds": 1.413960779998888,
+      "bytes": 11724293,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.2958799169973645,
-      "bytes": 6217387,
-      "pages": 35,
+      "seconds": 1.146538889999647,
+      "bytes": 5754870,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 0.9571121369990578,
-      "bytes": 4079115,
-      "pages": 20,
-      "shards": 2
+      "seconds": 1.1586719139995694,
+      "bytes": 7299186,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.3280163199997332,
-      "bytes": 13612044,
-      "pages": 47,
-      "shards": 4
+      "seconds": 3.9096268039975257,
+      "bytes": 6182999,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.827956348999578,
-      "bytes": 1165763,
+      "seconds": 0.8964433900000586,
+      "bytes": 1180194,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.215623644999141,
-      "bytes": 7281440,
-      "pages": 29,
+      "seconds": 1.5248420139978407,
+      "bytes": 7676145,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1079499599982228,
-      "bytes": 10281570,
-      "pages": 37,
+      "seconds": 1.2521359960010159,
+      "bytes": 10213095,
+      "pages": 25,
       "shards": 3
     },
     {
-      "seconds": 0.9046422840001469,
-      "bytes": 1080144,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.2234030319996236,
+      "bytes": 7751215,
+      "pages": 43,
+      "shards": 3
     },
     {
-      "seconds": 1.0989511400002812,
-      "bytes": 7379943,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.109631940998952,
+      "bytes": 6080871,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.0861735590005992,
-      "bytes": 6754175,
-      "pages": 31,
+      "seconds": 1.1276710530000855,
+      "bytes": 5237494,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2741842929972336,
-      "bytes": 12412930,
+      "seconds": 0.9725549989998399,
+      "bytes": 7624584,
       "pages": 39,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.0742992420018709,
-      "bytes": 6000047,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2354103249999753,
+      "bytes": 6135725,
+      "pages": 27,
+      "shards": 3
     },
     {
-      "seconds": 1.043194013000175,
-      "bytes": 771728,
-      "pages": 25,
+      "seconds": 1.0262013009996735,
+      "bytes": 964334,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 2.6395804720013984,
-      "bytes": 7630392,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.7832624309994571,
+      "bytes": 17985234,
+      "pages": 39,
+      "shards": 5
     },
     {
-      "seconds": 0.8558966809978301,
-      "bytes": 4770134,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.969149242999265,
+      "bytes": 1068713,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.0199386870008311,
-      "bytes": 6036267,
-      "pages": 31,
-      "shards": 2
+      "seconds": 1.4721264299987524,
+      "bytes": 7499280,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.4371028810019197,
-      "bytes": 13333064,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.0613430230005179,
+      "bytes": 6164642,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 0.925103517001844,
-      "bytes": 1157884,
-      "pages": 34,
+      "seconds": 0.908005076002155,
+      "bytes": 1082411,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.3517716829992423,
-      "bytes": 11962174,
-      "pages": 33,
+      "seconds": 27.483964940998703,
+      "bytes": 11288612,
+      "pages": 32,
       "shards": 4
     },
     {
-      "seconds": 0.8873195009982737,
-      "bytes": 1168035,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.235029557999951,
+      "bytes": 12065385,
+      "pages": 45,
+      "shards": 3
     },
     {
-      "seconds": 1.1477944020007271,
-      "bytes": 6045290,
-      "pages": 35,
-      "shards": 3
+      "seconds": 0.9373754530024598,
+      "bytes": 2569463,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.1455497599999944,
-      "bytes": 7253423,
-      "pages": 30,
+      "seconds": 1.3118033549981192,
+      "bytes": 11004186,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0725273009993543,
-      "bytes": 11023216,
-      "pages": 36,
-      "shards": 3
+      "seconds": 0.8848312870031805,
+      "bytes": 821123,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 2.0963671690005867,
-      "bytes": 7398466,
-      "pages": 37,
+      "seconds": 1.1538495610002428,
+      "bytes": 7799438,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0474640769971302,
-      "bytes": 900238,
-      "pages": 26,
-      "shards": 1
+      "seconds": 0.8605145479996281,
+      "bytes": 5751347,
+      "pages": 25,
+      "shards": 2
     },
     {
-      "seconds": 1.1159279130006325,
-      "bytes": 6352582,
-      "pages": 36,
+      "seconds": 1.0857413720004843,
+      "bytes": 7211440,
+      "pages": 43,
       "shards": 2
     },
     {
-      "seconds": 1.1081351329994504,
-      "bytes": 11588439,
-      "pages": 32,
+      "seconds": 1.2849263329990208,
+      "bytes": 11334677,
+      "pages": 35,
       "shards": 4
     },
     {
-      "seconds": 0.8398587470001075,
-      "bytes": 1103964,
-      "pages": 34,
+      "seconds": 1.0467089929989015,
+      "bytes": 955823,
+      "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.1820991080021486,
-      "bytes": 11679771,
+      "seconds": 1.141221003003011,
+      "bytes": 7503350,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1287752040007035,
+      "bytes": 10564318,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.2690833449996717,
-      "bytes": 2379532,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.0117927669998608,
+      "bytes": 847459,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.1314427079996676,
-      "bytes": 10762955,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.3560152030004247,
+      "bytes": 13557643,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.0500791910017142,
-      "bytes": 5691792,
+      "seconds": 1.124354845000198,
+      "bytes": 904261,
       "pages": 27,
-      "shards": 2
+      "shards": 1
     },
     {
-      "seconds": 1.035739069000556,
-      "bytes": 2751840,
-      "pages": 37,
+      "seconds": 1.0335092330024054,
+      "bytes": 5888838,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.1663796350003395,
-      "bytes": 6310887,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2323054039989074,
+      "bytes": 11963716,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.171979762999399,
-      "bytes": 12042987,
-      "pages": 27,
-      "shards": 4
+      "seconds": 0.7193277299993497,
+      "bytes": 1174751,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.1372878729998774,
-      "bytes": 6231473,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.1924678350005706,
+      "bytes": 7370627,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 0.9854615350013773,
-      "bytes": 5689074,
-      "pages": 28,
+      "seconds": 0.999219997000182,
+      "bytes": 6670501,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.9606277580023743,
-      "bytes": 2676212,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.1155471289966954,
+      "bytes": 10846938,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.570391033001215,
-      "bytes": 9821415,
-      "pages": 34,
+      "seconds": 1.1072990959983144,
+      "bytes": 7478866,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1360236319997057,
-      "bytes": 2731200,
+      "seconds": 0.9801313110001502,
+      "bytes": 949416,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2882471760021872,
+      "bytes": 6045719,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.413960779998888,
-      "bytes": 11724293,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.362017162002303,
+      "bytes": 11390848,
+      "pages": 38,
+      "shards": 5
     },
     {
-      "seconds": 1.146538889999647,
-      "bytes": 5754870,
-      "pages": 32,
+      "seconds": 0.9684943469983409,
+      "bytes": 6504817,
+      "pages": 23,
       "shards": 2
     },
     {
-      "seconds": 1.1586719139995694,
-      "bytes": 7299186,
-      "pages": 37,
+      "seconds": 1.0333191259996966,
+      "bytes": 7473380,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 3.9096268039975257,
-      "bytes": 6182999,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.8964433900000586,
-      "bytes": 1180194,
-      "pages": 34,
+      "seconds": 0.8470484819990816,
+      "bytes": 1045811,
+      "pages": 30,
       "shards": 1
     },
     {
-      "seconds": 1.5248420139978407,
-      "bytes": 7676145,
-      "pages": 37,
+      "seconds": 1.2339288350012794,
+      "bytes": 10876745,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.2521359960010159,
-      "bytes": 10213095,
-      "pages": 25,
-      "shards": 3
+      "seconds": 1.0449587049988622,
+      "bytes": 2671135,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.2234030319996236,
-      "bytes": 7751215,
-      "pages": 43,
-      "shards": 3
+      "seconds": 1.0467685680014256,
+      "bytes": 5999640,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 1.109631940998952,
-      "bytes": 6080871,
-      "pages": 36,
+      "seconds": 1.1004228650017467,
+      "bytes": 6536803,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.1276710530000855,
-      "bytes": 5237494,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.3272996070008958,
+      "bytes": 12496214,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 0.9725549989998399,
-      "bytes": 7624584,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.0267054780015314,
+      "bytes": 5908271,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.2354103249999753,
-      "bytes": 6135725,
-      "pages": 27,
-      "shards": 3
+      "seconds": 0.8663640420018055,
+      "bytes": 2740775,
+      "pages": 36,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dee880db6477edfe4f23e2084b87d73ef0b8e70abdddbdfbbee6e8595a7fc54b
-size 721778047
+oid sha256:b48df57d11cb4744bb19a078c1bcb987e665797abd0171d45df1688e70bd7bcf
+size 721783231