Qwen90M step 350000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 345000,
+  "step": 350000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "1280a8081ba139ed35dc86617f2d00c744f61dc8d28e7148fdcf6e8ecb03db3e",
+    "model.safetensors": "f5426f370d1251d7bba5de89b56b33e67df71399f079e2eeb4e3700f38dc4b31",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "b48df57d11cb4744bb19a078c1bcb987e665797abd0171d45df1688e70bd7bcf",
+    "trainer.pt": "b2c2d3d080843d366ffbe0938a83b313ceff09a34962f3581968203e2d4b079d",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "90ad39d684ac792f235cbfbeef498d4c81f09977d8f89fb8c3bdd41e515f347a",
-    "shard_scheduler.json": "ed5ff406bb283085db0d0d6ae907c0411de8aeb884c3d97dca7e818c32facac2",
-    "coverage.json": "2b001d842650d60a89eccf53c95a56122f777cfa328d77fd27f8403912f812c5",
+    "shard_scheduler.json": "fad92bf0a8e73e7059f462c17088165edab733c4deb3ec42d95e84132126c970",
+    "coverage.json": "c9baf1742338a56c30b328003396bc082f4bfbced4f8bd6b990d6a8a6cef1d28",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 345000,
+  "step": 350000,
   "stats": {
-    "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
+    "climbmix/blocks": 12600000,
+    "climbmix/targets": 21063108714,
+    "climbmix/input_tokens": 21064443608,
+    "ultra_qa/blocks": 4725000,
+    "ultra_qa/targets": 7708127333,
+    "ultra_qa/input_tokens": 7708142241,
+    "cortex/blocks": 4725000,
+    "cortex/targets": 179755192,
+    "cortex/input_tokens": 7511179034,
+    "ultra_style/blocks": 4725000,
+    "ultra_style/targets": 7473844592,
+    "ultra_style/input_tokens": 7474236913,
+    "openmath/blocks": 3150000,
+    "openmath/targets": 4867680878,
+    "openmath/input_tokens": 4869303434,
+    "rewrite6/blocks": 1575000,
+    "rewrite6/targets": 2581925316,
+    "rewrite6/input_tokens": 2584756299
   },
   "sources": {
     "climbmix": {
-      "page": 1701,
-      "row": 907,
+      "page": 1725,
+      "row": 3105,
       "start": {
-        "file": 82,
+        "file": 83,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 27,
+          "row_group": 41,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 6968203
+      "documents": 7068705
     },
     "rewrite6": {
-      "page": 17,
-      "row": 3186,
+      "page": 31,
+      "row": 3505,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 69632
+          "row": 126976
         }
       },
       "epoch": 19,
-      "offset": 2048,
-      "documents": 3975494
+      "offset": 0,
+      "documents": 4033157
     },
     "openmath": {
-      "page": 1063,
-      "row": 868,
+      "page": 1078,
+      "row": 2678,
       "start": {
-        "file": 9,
+        "file": 10,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 424,
-          "row": 198
+          "row_group": 48,
+          "row": 988
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 4354916
+      "documents": 4418166
     },
     "ultra_qa": {
-      "page": 459,
-      "row": 2604,
+      "page": 466,
+      "row": 1192,
       "start": {
         "file": 3,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 4,
-          "row": 62811
+          "row_group": 5,
+          "row": 26875
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 1882668
+      "offset": 2048,
+      "documents": 1909928
     },
     "ultra_style": {
-      "page": 701,
-      "row": 2909,
+      "page": 711,
+      "row": 3707,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 1,
-          "row": 46070
+          "row_group": 2,
+          "row": 2093
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 2874205
+      "documents": 2915963
     },
     "cortex": {
-      "page": 90333,
-      "row": 88,
+      "page": 91642,
+      "row": 117,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 11562712
+      "documents": 11730293
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"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"}
+{"step": 350000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "f5426f370d1251d7bba5de89b56b33e67df71399f079e2eeb4e3700f38dc4b31", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "b2c2d3d080843d366ffbe0938a83b313ceff09a34962f3581968203e2d4b079d", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "90ad39d684ac792f235cbfbeef498d4c81f09977d8f89fb8c3bdd41e515f347a", "shard_scheduler.json": "fad92bf0a8e73e7059f462c17088165edab733c4deb3ec42d95e84132126c970", "coverage.json": "c9baf1742338a56c30b328003396bc082f4bfbced4f8bd6b990d6a8a6cef1d28", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000350000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1280a8081ba139ed35dc86617f2d00c744f61dc8d28e7148fdcf6e8ecb03db3e
+oid sha256:f5426f370d1251d7bba5de89b56b33e67df71399f079e2eeb4e3700f38dc4b31
 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": 75294.19321551842,
-      "required_pages_per_second_upper_estimate": 0.006526309836533097
+      "required_token_positions_per_second": 75583.06335444357,
+      "required_pages_per_second_upper_estimate": 0.006551348368040456
     },
     {
       "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": 9411.774151939802,
-      "required_pages_per_second_upper_estimate": 0.00375449742777238
+      "required_token_positions_per_second": 9447.882919305446,
+      "required_pages_per_second_upper_estimate": 0.003768901754948718
     },
     {
       "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": 18823.548303879605,
-      "required_pages_per_second_upper_estimate": 0.004354496616967123
+      "required_token_positions_per_second": 18895.765838610892,
+      "required_pages_per_second_upper_estimate": 0.004371202872642004
     },
     {
       "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": 28235.32245581941,
-      "required_pages_per_second_upper_estimate": 0.0017193829982712838
+      "required_token_positions_per_second": 28343.648757916337,
+      "required_pages_per_second_upper_estimate": 0.0017259795017248035
     },
     {
       "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": 28235.32245581941,
-      "required_pages_per_second_upper_estimate": 0.0021736332830240225
+      "required_token_positions_per_second": 28343.648757916337,
+      "required_pages_per_second_upper_estimate": 0.0021819725416258404
     },
     {
       "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": 28235.32245581941,
-      "required_pages_per_second_upper_estimate": 0.3505882073558664
+      "required_token_positions_per_second": 28343.648757916337,
+      "required_pages_per_second_upper_estimate": 0.3519332574830989
     }
   ],
   "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": 0.11289917382015346,
-    "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,
+    "producer_batch_seconds": 0.11979209478852226,
+    "producer_p95_seconds": 0.38984344500204315,
+    "gpu_update_seconds": 0.975456626496543,
+    "producer_batches_per_second": 8.34779625287773,
+    "gpu_batches_per_second": 1.0251609070426915,
+    "producer_headroom": 8.142913171512593,
     "sustainable": true,
-    "buffer_seconds": 60.71033906845649,
+    "buffer_seconds": 60.47831084278566,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
     "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 312717.6648123108,
-    "upload_bytes_per_second": 4152064.367452919,
+    "generation_bytes_per_second": 379734.8500496105,
+    "upload_bytes_per_second": 4092962.040476732,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 67.10193117405885,
+    "estimated_full_shards_per_hour": 81.48225904575573,
     "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": 1.2690833449996717,
-      "bytes": 2379532,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.1287752040007035,
+      "bytes": 10564318,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.1314427079996676,
-      "bytes": 10762955,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.0117927669998608,
+      "bytes": 847459,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.0500791910017142,
-      "bytes": 5691792,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.3560152030004247,
+      "bytes": 13557643,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.035739069000556,
-      "bytes": 2751840,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.124354845000198,
+      "bytes": 904261,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.1663796350003395,
-      "bytes": 6310887,
-      "pages": 36,
+      "seconds": 1.0335092330024054,
+      "bytes": 5888838,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.171979762999399,
-      "bytes": 12042987,
-      "pages": 27,
+      "seconds": 1.2323054039989074,
+      "bytes": 11963716,
+      "pages": 40,
       "shards": 4
     },
     {
-      "seconds": 1.1372878729998774,
-      "bytes": 6231473,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.7193277299993497,
+      "bytes": 1174751,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 0.9854615350013773,
-      "bytes": 5689074,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1924678350005706,
+      "bytes": 7370627,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 0.9606277580023743,
-      "bytes": 2676212,
-      "pages": 36,
+      "seconds": 0.999219997000182,
+      "bytes": 6670501,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.570391033001215,
-      "bytes": 9821415,
-      "pages": 34,
+      "seconds": 1.1155471289966954,
+      "bytes": 10846938,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.1360236319997057,
-      "bytes": 2731200,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.1072990959983144,
+      "bytes": 7478866,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.413960779998888,
-      "bytes": 11724293,
-      "pages": 29,
-      "shards": 3
+      "seconds": 0.9801313110001502,
+      "bytes": 949416,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.146538889999647,
-      "bytes": 5754870,
-      "pages": 32,
+      "seconds": 1.2882471760021872,
+      "bytes": 6045719,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1586719139995694,
-      "bytes": 7299186,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.362017162002303,
+      "bytes": 11390848,
+      "pages": 38,
+      "shards": 5
     },
     {
-      "seconds": 3.9096268039975257,
-      "bytes": 6182999,
-      "pages": 28,
+      "seconds": 0.9684943469983409,
+      "bytes": 6504817,
+      "pages": 23,
       "shards": 2
     },
     {
-      "seconds": 0.8964433900000586,
-      "bytes": 1180194,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.0333191259996966,
+      "bytes": 7473380,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.5248420139978407,
-      "bytes": 7676145,
-      "pages": 37,
-      "shards": 3
+      "seconds": 0.8470484819990816,
+      "bytes": 1045811,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 1.2521359960010159,
-      "bytes": 10213095,
-      "pages": 25,
+      "seconds": 1.2339288350012794,
+      "bytes": 10876745,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.2234030319996236,
-      "bytes": 7751215,
-      "pages": 43,
-      "shards": 3
+      "seconds": 1.0449587049988622,
+      "bytes": 2671135,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.109631940998952,
-      "bytes": 6080871,
-      "pages": 36,
+      "seconds": 1.0467685680014256,
+      "bytes": 5999640,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.1276710530000855,
-      "bytes": 5237494,
-      "pages": 29,
+      "seconds": 1.1004228650017467,
+      "bytes": 6536803,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 0.9725549989998399,
-      "bytes": 7624584,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.3272996070008958,
+      "bytes": 12496214,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 1.2354103249999753,
-      "bytes": 6135725,
-      "pages": 27,
-      "shards": 3
+      "seconds": 1.0267054780015314,
+      "bytes": 5908271,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.0262013009996735,
-      "bytes": 964334,
-      "pages": 29,
-      "shards": 1
+      "seconds": 0.8663640420018055,
+      "bytes": 2740775,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.7832624309994571,
-      "bytes": 17985234,
-      "pages": 39,
-      "shards": 5
+      "seconds": 0.9017275879996305,
+      "bytes": 5703769,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.969149242999265,
-      "bytes": 1068713,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.2448231849994045,
+      "bytes": 5850047,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.4721264299987524,
-      "bytes": 7499280,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.1455546180004603,
+      "bytes": 12663721,
+      "pages": 38,
+      "shards": 4
     },
     {
-      "seconds": 1.0613430230005179,
-      "bytes": 6164642,
-      "pages": 29,
+      "seconds": 1.2500024359978852,
+      "bytes": 6221044,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.908005076002155,
-      "bytes": 1082411,
+      "seconds": 0.9838440579987946,
+      "bytes": 1078556,
       "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 27.483964940998703,
-      "bytes": 11288612,
-      "pages": 32,
-      "shards": 4
-    },
-    {
-      "seconds": 1.235029557999951,
-      "bytes": 12065385,
-      "pages": 45,
+      "seconds": 1.215074095998716,
+      "bytes": 7808319,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 0.9373754530024598,
-      "bytes": 2569463,
+      "seconds": 1.0071898829992278,
+      "bytes": 5466457,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.3118033549981192,
-      "bytes": 11004186,
-      "pages": 34,
+      "seconds": 1.1624265700011165,
+      "bytes": 7895099,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.8848312870031805,
-      "bytes": 821123,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.032417123002233,
+      "bytes": 5759947,
+      "pages": 28,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9372151059978933,
+      "bytes": 7005197,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1538495610002428,
-      "bytes": 7799438,
+      "seconds": 34.0496005649984,
+      "bytes": 11526687,
       "pages": 38,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 0.8605145479996281,
-      "bytes": 5751347,
-      "pages": 25,
+      "seconds": 1.0691167909972137,
+      "bytes": 6134154,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0857413720004843,
-      "bytes": 7211440,
-      "pages": 43,
+      "seconds": 1.9025495240020973,
+      "bytes": 2606249,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2849263329990208,
-      "bytes": 11334677,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.0264649749988166,
+      "bytes": 10585305,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.0467089929989015,
-      "bytes": 955823,
-      "pages": 27,
+      "seconds": 0.7573466569992888,
+      "bytes": 999325,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.141221003003011,
-      "bytes": 7503350,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.4190698960010195,
+      "bytes": 13694673,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.1287752040007035,
-      "bytes": 10564318,
-      "pages": 35,
-      "shards": 3
+      "seconds": 0.8739250739999989,
+      "bytes": 5944089,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.0117927669998608,
-      "bytes": 847459,
-      "pages": 26,
+      "seconds": 0.9463060489979398,
+      "bytes": 743440,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 1.3560152030004247,
-      "bytes": 13557643,
-      "pages": 39,
+      "seconds": 1.0638403249977273,
+      "bytes": 11964464,
+      "pages": 46,
       "shards": 4
     },
     {
-      "seconds": 1.124354845000198,
-      "bytes": 904261,
+      "seconds": 1.8820245999959297,
+      "bytes": 865291,
       "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.0335092330024054,
-      "bytes": 5888838,
-      "pages": 31,
-      "shards": 2
+      "seconds": 1.3777121140010422,
+      "bytes": 7459243,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.2323054039989074,
-      "bytes": 11963716,
-      "pages": 40,
-      "shards": 4
+      "seconds": 2.28389226100262,
+      "bytes": 15733663,
+      "pages": 34,
+      "shards": 5
     },
     {
-      "seconds": 0.7193277299993497,
-      "bytes": 1174751,
-      "pages": 35,
+      "seconds": 0.7980627049983013,
+      "bytes": 1094775,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.1924678350005706,
-      "bytes": 7370627,
-      "pages": 30,
+      "seconds": 1.0217303430035827,
+      "bytes": 7986291,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.999219997000182,
-      "bytes": 6670501,
-      "pages": 27,
-      "shards": 2
+      "seconds": 0.8759991519982577,
+      "bytes": 805335,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.1155471289966954,
-      "bytes": 10846938,
-      "pages": 39,
+      "seconds": 1.0561498970055254,
+      "bytes": 10650758,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.1072990959983144,
-      "bytes": 7478866,
+      "seconds": 1.2883731410038308,
+      "bytes": 7659801,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9801313110001502,
-      "bytes": 949416,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.1231011929994565,
+      "bytes": 6897324,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.2882471760021872,
-      "bytes": 6045719,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.0080412499955855,
+      "bytes": 7466769,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.362017162002303,
-      "bytes": 11390848,
-      "pages": 38,
-      "shards": 5
+      "seconds": 1.1737449919965002,
+      "bytes": 4963940,
+      "pages": 26,
+      "shards": 2
     },
     {
-      "seconds": 0.9684943469983409,
-      "bytes": 6504817,
-      "pages": 23,
+      "seconds": 1.0047582629995304,
+      "bytes": 6204635,
+      "pages": 42,
       "shards": 2
     },
     {
-      "seconds": 1.0333191259996966,
-      "bytes": 7473380,
-      "pages": 38,
+      "seconds": 1.2032458940011566,
+      "bytes": 7462591,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 0.8470484819990816,
-      "bytes": 1045811,
-      "pages": 30,
+      "seconds": 1.0979009180009598,
+      "bytes": 1059989,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.2339288350012794,
-      "bytes": 10876745,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.474424235006154,
+      "bytes": 14927789,
+      "pages": 21,
+      "shards": 4
     },
     {
-      "seconds": 1.0449587049988622,
-      "bytes": 2671135,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.0885418239995488,
+      "bytes": 8199005,
+      "pages": 47,
+      "shards": 3
     },
     {
-      "seconds": 1.0467685680014256,
-      "bytes": 5999640,
-      "pages": 31,
-      "shards": 2
+      "seconds": 0.9340340310009196,
+      "bytes": 958542,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 1.1004228650017467,
-      "bytes": 6536803,
-      "pages": 24,
-      "shards": 2
+      "seconds": 1.0266037770052208,
+      "bytes": 7314912,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.3272996070008958,
-      "bytes": 12496214,
-      "pages": 42,
-      "shards": 4
+      "seconds": 1.1214634670031955,
+      "bytes": 5616787,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.0267054780015314,
-      "bytes": 5908271,
-      "pages": 28,
+      "seconds": 1.0272974420004175,
+      "bytes": 6127988,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8663640420018055,
-      "bytes": 2740775,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.1608434069930809,
+      "bytes": 13413721,
+      "pages": 38,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b48df57d11cb4744bb19a078c1bcb987e665797abd0171d45df1688e70bd7bcf
-size 721783231
+oid sha256:b2c2d3d080843d366ffbe0938a83b313ceff09a34962f3581968203e2d4b079d
+size 721769855