Qwen90M step 15000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 10000,
+  "step": 15000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "dbbe11b6b37a2fb226a0278df00d43ccaefb62a99900fed5fe3ea03c03c5f9a7",
+    "model.safetensors": "cd4b4a336ca854cc63b56daafd448bbdabc975be153afb5b40dbcfeed84c0a5c",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "809e3a6fcad791a5e8798c9628745256615bd83b94d0bf2a79256dfb46e2c777",
+    "trainer.pt": "9ee3cdbc8544ee947af4a7d860b9cd7d6a5386d7eb87fa5e677065b54efe108a",
     "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2",
     "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e",
-    "shard_scheduler.json": "0b2e87c4e5b8348b75947413ade06d08fb96284525563d4f5bff035e5155a90e",
-    "coverage.json": "f1c3b8c37327537faadaa6ef151f85ed40e99aa77fa55c6977cf180d61920dd1",
+    "shard_scheduler.json": "4d92f7ac5793b7b10fc6dcd78ea5d5fc6a289ca95e0d0b954b4effa3df49ceff",
+    "coverage.json": "ae6b95743a2e517acaa8daca9ca6ec0408a66feeeb897332ac3aa2b7585e4311",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 10000,
+  "step": 15000,
   "stats": {
-    "climbmix/blocks": 360000,
-    "climbmix/targets": 600233197,
-    "climbmix/input_tokens": 600271805,
-    "ultra_qa/blocks": 135000,
-    "ultra_qa/targets": 220395872,
-    "ultra_qa/input_tokens": 220396305,
-    "cortex/blocks": 135000,
-    "cortex/targets": 5141709,
-    "cortex/input_tokens": 214618986,
-    "ultra_style/blocks": 135000,
-    "ultra_style/targets": 213420214,
-    "ultra_style/input_tokens": 213431604,
-    "openmath/blocks": 90000,
-    "openmath/targets": 139182081,
-    "openmath/input_tokens": 139228806,
-    "rewrite6/blocks": 45000,
-    "rewrite6/targets": 73701578,
-    "rewrite6/input_tokens": 73782727
+    "climbmix/blocks": 540000,
+    "climbmix/targets": 901514897,
+    "climbmix/input_tokens": 901572469,
+    "ultra_qa/blocks": 202500,
+    "ultra_qa/targets": 330561975,
+    "ultra_qa/input_tokens": 330562626,
+    "cortex/blocks": 202500,
+    "cortex/targets": 7710129,
+    "cortex/input_tokens": 321940622,
+    "ultra_style/blocks": 202500,
+    "ultra_style/targets": 320384397,
+    "ultra_style/input_tokens": 320401468,
+    "openmath/blocks": 135000,
+    "openmath/targets": 208599081,
+    "openmath/input_tokens": 208668782,
+    "rewrite6/blocks": 67500,
+    "rewrite6/targets": 110633666,
+    "rewrite6/input_tokens": 110755126
   },
   "sources": {
     "climbmix": {
-      "page": 49,
-      "row": 3403,
+      "page": 74,
+      "row": 1396,
       "start": {
-        "file": 2,
+        "file": 3,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 27,
+          "row_group": 44,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 204107
+      "documents": 304500
     },
     "rewrite6": {
-      "page": 28,
-      "row": 858,
+      "page": 42,
+      "row": 1009,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 114688
+          "row": 172032
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 115546
+      "documents": 173041
     },
     "openmath": {
-      "page": 30,
-      "row": 3710,
+      "page": 46,
+      "row": 1133,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 122,
-          "row": 880
+          "row_group": 188,
+          "row": 416
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 126590
+      "documents": 189549
     },
     "ultra_qa": {
-      "page": 13,
-      "row": 1338,
+      "page": 20,
+      "row": 1,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 0,
-          "row": 53248
+          "row_group": 1,
+          "row": 17350
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 54586
+      "offset": 4096,
+      "documents": 81921
     },
     "ultra_style": {
-      "page": 20,
-      "row": 1825,
+      "page": 30,
+      "row": 2561,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 0,
-          "row": 81920
+          "row_group": 1,
+          "row": 37513
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 83745
+      "documents": 125441
     },
     "cortex": {
-      "page": 2617,
-      "row": 78,
+      "page": 3926,
+      "row": 10,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 335054
+      "documents": 502538
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 10000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "dbbe11b6b37a2fb226a0278df00d43ccaefb62a99900fed5fe3ea03c03c5f9a7", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "809e3a6fcad791a5e8798c9628745256615bd83b94d0bf2a79256dfb46e2c777", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "0b2e87c4e5b8348b75947413ade06d08fb96284525563d4f5bff035e5155a90e", "coverage.json": "f1c3b8c37327537faadaa6ef151f85ed40e99aa77fa55c6977cf180d61920dd1", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000010000"}
+{"step": 15000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "cd4b4a336ca854cc63b56daafd448bbdabc975be153afb5b40dbcfeed84c0a5c", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "9ee3cdbc8544ee947af4a7d860b9cd7d6a5386d7eb87fa5e677065b54efe108a", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "4d92f7ac5793b7b10fc6dcd78ea5d5fc6a289ca95e0d0b954b4effa3df49ceff", "coverage.json": "ae6b95743a2e517acaa8daca9ca6ec0408a66feeeb897332ac3aa2b7585e4311", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000015000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dbbe11b6b37a2fb226a0278df00d43ccaefb62a99900fed5fe3ea03c03c5f9a7
+oid sha256:cd4b4a336ca854cc63b56daafd448bbdabc975be153afb5b40dbcfeed84c0a5c
 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": 75436.00239654537,
-      "required_pages_per_second_upper_estimate": 0.00653860149693243
+      "required_token_positions_per_second": 75908.27937948794,
+      "required_pages_per_second_upper_estimate": 0.006579537268838822
     },
     {
       "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": 9429.500299568172,
-      "required_pages_per_second_upper_estimate": 0.0037615686530908616
+      "required_token_positions_per_second": 9488.534922435992,
+      "required_pages_per_second_upper_estimate": 0.0037851184467991034
     },
     {
       "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": 18859.000599136343,
-      "required_pages_per_second_upper_estimate": 0.004362697881535683
+      "required_token_positions_per_second": 18977.069844871985,
+      "required_pages_per_second_upper_estimate": 0.004390011123588869
     },
     {
       "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": 28288.500898704515,
-      "required_pages_per_second_upper_estimate": 0.0017226212864372597
+      "required_token_positions_per_second": 28465.604767307974,
+      "required_pages_per_second_upper_estimate": 0.0017334059828430235
     },
     {
       "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": 28288.500898704515,
-      "required_pages_per_second_upper_estimate": 0.0021777271067646695
+      "required_token_positions_per_second": 28465.604767307974,
+      "required_pages_per_second_upper_estimate": 0.0021913610528246498
     },
     {
       "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": 28288.500898704515,
-      "required_pages_per_second_upper_estimate": 0.35124850563969995
+      "required_token_positions_per_second": 28465.604767307974,
+      "required_pages_per_second_upper_estimate": 0.3534475429592358
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.38050877037498765,
-    "producer_p95_seconds": 0.2422439990004932,
-    "gpu_update_seconds": 0.977358259421451,
-    "producer_batches_per_second": 2.628060317806893,
-    "gpu_batches_per_second": 1.023166265144116,
-    "producer_headroom": 2.5685564578663302,
+    "producer_batch_seconds": 0.6275874297735697,
+    "producer_p95_seconds": 0.29346390000137035,
+    "gpu_update_seconds": 0.9712774496101002,
+    "producer_batches_per_second": 1.5934034885956763,
+    "gpu_batches_per_second": 1.029571931687933,
+    "producer_headroom": 1.5476368766030448,
     "sustainable": true,
-    "buffer_seconds": 60.59621208412997,
+    "buffer_seconds": 60.219201875826215,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
     "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 187927.3360595079,
-    "upload_bytes_per_second": 6523426.966652929,
+    "generation_bytes_per_second": 234605.4285099929,
+    "upload_bytes_per_second": 6302700.702389904,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 40.32483159388474,
+    "estimated_full_shards_per_hour": 50.340863623379136,
     "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": 0.8951566169998841,
-      "bytes": 8341517,
-      "pages": 32,
-      "shards": 3
-    },
-    {
-      "seconds": 1.104310605000137,
-      "bytes": 5895784,
-      "pages": 34,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2573788659997263,
-      "bytes": 8080667,
-      "pages": 37,
+      "seconds": 1.0970937430001868,
+      "bytes": 8464356,
+      "pages": 36,
       "shards": 4
     },
     {
-      "seconds": 0.9144526599998244,
-      "bytes": 5532432,
-      "pages": 26,
+      "seconds": 1.18106462400101,
+      "bytes": 6740941,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.9253262060001362,
-      "bytes": 5975019,
+      "seconds": 1.0259071060008864,
+      "bytes": 6033717,
       "pages": 35,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 0.8811729529998047,
-      "bytes": 3757694,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.3882136779993743,
+      "bytes": 13040894,
+      "pages": 40,
+      "shards": 5
     },
     {
-      "seconds": 1.5062500680000994,
-      "bytes": 15040116,
-      "pages": 34,
-      "shards": 4
+      "seconds": 0.7154928489999293,
+      "bytes": 835967,
+      "pages": 25,
+      "shards": 1
     },
     {
-      "seconds": 1.0638150449999557,
-      "bytes": 7164962,
-      "pages": 39,
+      "seconds": 1.0692436869994708,
+      "bytes": 7577178,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.0547767480002221,
-      "bytes": 2359076,
-      "pages": 27,
+      "seconds": 1.0859860309992655,
+      "bytes": 2128705,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.027329577999808,
-      "bytes": 11020380,
-      "pages": 34,
+      "seconds": 1.1122738659996685,
+      "bytes": 11663489,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.2178119280006285,
-      "bytes": 3672261,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.1172626269999455,
+      "bytes": 11245903,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.0616448490000039,
-      "bytes": 5911843,
-      "pages": 30,
+      "seconds": 0.9253301299995655,
+      "bytes": 2328336,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.2120231849994525,
-      "bytes": 11658602,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9783495700003186,
-      "bytes": 7342912,
-      "pages": 40,
-      "shards": 4
-    },
-    {
-      "seconds": 0.9914971539992621,
-      "bytes": 6238955,
+      "seconds": 1.028129626000009,
+      "bytes": 6087526,
       "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9600033409997195,
-      "bytes": 1092240,
-      "pages": 31,
-      "shards": 1
+      "seconds": 0.7597279170004185,
+      "bytes": 2156570,
+      "pages": 21,
+      "shards": 2
     },
     {
-      "seconds": 0.9772284549999313,
-      "bytes": 8526326,
-      "pages": 38,
+      "seconds": 0.9871070610006427,
+      "bytes": 12008396,
+      "pages": 45,
       "shards": 4
     },
     {
-      "seconds": 1.260473302999344,
-      "bytes": 10994776,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.228597503999481,
+      "bytes": 13077636,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 0.9115331740003967,
-      "bytes": 8005286,
-      "pages": 27,
-      "shards": 3
+      "seconds": 1.1013024290004978,
+      "bytes": 1132593,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 0.9674303089996101,
-      "bytes": 7444249,
-      "pages": 41,
+      "seconds": 0.9999866519992793,
+      "bytes": 6918002,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 0.8210835840000072,
-      "bytes": 900296,
-      "pages": 25,
-      "shards": 1
-    },
-    {
-      "seconds": 1.093376490999617,
-      "bytes": 11563576,
+      "seconds": 1.2065209279990086,
+      "bytes": 11423370,
       "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 0.9312379190005231,
-      "bytes": 7230860,
-      "pages": 33,
+      "seconds": 1.0073676210013218,
+      "bytes": 932900,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.0279631710000103,
+      "bytes": 7068347,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.8876133860003392,
-      "bytes": 811866,
-      "pages": 25,
-      "shards": 1
+      "seconds": 0.8251235360003193,
+      "bytes": 2046300,
+      "pages": 19,
+      "shards": 2
     },
     {
-      "seconds": 1.1758913479998228,
-      "bytes": 13372381,
-      "pages": 39,
+      "seconds": 1.6571312910000415,
+      "bytes": 16761254,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 1.208357038000031,
-      "bytes": 5841536,
-      "pages": 34,
-      "shards": 3
+      "seconds": 0.7977973630004271,
+      "bytes": 2044807,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.0383053309997194,
-      "bytes": 7595630,
-      "pages": 37,
+      "seconds": 1.0466349409998656,
+      "bytes": 7179273,
+      "pages": 20,
       "shards": 3
     },
     {
-      "seconds": 1.335573416000443,
-      "bytes": 5731224,
-      "pages": 27,
-      "shards": 2
+      "seconds": 0.7437304710001627,
+      "bytes": 927772,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.0351095349997195,
-      "bytes": 2158209,
-      "pages": 32,
+      "seconds": 1.1199327470003482,
+      "bytes": 6072182,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.3273477019993152,
-      "bytes": 12043048,
-      "pages": 41,
+      "seconds": 0.9855451060011546,
+      "bytes": 8735177,
+      "pages": 38,
       "shards": 4
     },
     {
-      "seconds": 0.7578286200005095,
-      "bytes": 992201,
-      "pages": 26,
-      "shards": 1
+      "seconds": 0.964613582000311,
+      "bytes": 5800981,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.176080047999676,
-      "bytes": 12981710,
+      "seconds": 0.8288159710009495,
+      "bytes": 7059901,
       "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9880495930010511,
+      "bytes": 8851756,
+      "pages": 43,
       "shards": 4
     },
     {
-      "seconds": 1.0470021729997825,
-      "bytes": 7549840,
-      "pages": 34,
+      "seconds": 1.0367009409983439,
+      "bytes": 10231090,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.1135005340001953,
-      "bytes": 5221142,
-      "pages": 32,
-      "shards": 2
+      "seconds": 0.8111973979994218,
+      "bytes": 875148,
+      "pages": 25,
+      "shards": 1
     },
     {
-      "seconds": 1.3410022780008148,
-      "bytes": 8747781,
-      "pages": 41,
+      "seconds": 1.2982671780009696,
+      "bytes": 8839899,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 0.8860283060002985,
-      "bytes": 800987,
-      "pages": 26,
+      "seconds": 0.756701359001454,
+      "bytes": 1071685,
+      "pages": 31,
       "shards": 1
     },
     {
-      "seconds": 0.8824849620004898,
-      "bytes": 7297737,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.2869488490014191,
+      "bytes": 16651954,
+      "pages": 29,
+      "shards": 5
     },
     {
-      "seconds": 1.4247361650004677,
-      "bytes": 17857886,
-      "pages": 36,
-      "shards": 5
+      "seconds": 0.9651243050011544,
+      "bytes": 7745172,
+      "pages": 44,
+      "shards": 4
     },
     {
-      "seconds": 0.7635651720001988,
-      "bytes": 1246011,
-      "pages": 34,
+      "seconds": 0.7505128630000399,
+      "bytes": 1172556,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.6194143330003499,
-      "bytes": 8763049,
-      "pages": 35,
+      "seconds": 1.1141684859994712,
+      "bytes": 8597889,
+      "pages": 33,
       "shards": 4
     },
     {
-      "seconds": 0.7844010799999523,
-      "bytes": 865744,
+      "seconds": 1.1277135729997099,
+      "bytes": 5536865,
+      "pages": 31,
+      "shards": 2
+    },
+    {
+      "seconds": 0.9057988900003693,
+      "bytes": 5888539,
+      "pages": 27,
+      "shards": 2
+    },
+    {
+      "seconds": 1.030597670998759,
+      "bytes": 7980071,
       "pages": 26,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 1.143092911000167,
-      "bytes": 10241728,
-      "pages": 39,
+      "seconds": 0.9780282619994978,
+      "bytes": 7468851,
+      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 1.0970937430001868,
-      "bytes": 8464356,
-      "pages": 36,
+      "seconds": 1.20604977199946,
+      "bytes": 11385510,
+      "pages": 35,
       "shards": 4
     },
     {
-      "seconds": 1.18106462400101,
-      "bytes": 6740941,
-      "pages": 29,
-      "shards": 2
+      "seconds": 0.7792264220006473,
+      "bytes": 714539,
+      "pages": 19,
+      "shards": 1
     },
     {
-      "seconds": 1.0259071060008864,
-      "bytes": 6033717,
-      "pages": 35,
+      "seconds": 1.1150978799996665,
+      "bytes": 7568611,
+      "pages": 45,
       "shards": 3
     },
     {
-      "seconds": 1.3882136779993743,
-      "bytes": 13040894,
-      "pages": 40,
-      "shards": 5
+      "seconds": 1.1983624430013151,
+      "bytes": 7531256,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 0.7154928489999293,
-      "bytes": 835967,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.255419142999017,
+      "bytes": 11119971,
+      "pages": 21,
+      "shards": 3
     },
     {
-      "seconds": 1.0692436869994708,
-      "bytes": 7577178,
-      "pages": 36,
+      "seconds": 1.229843424000137,
+      "bytes": 7832261,
+      "pages": 45,
       "shards": 3
     },
     {
-      "seconds": 1.0859860309992655,
-      "bytes": 2128705,
-      "pages": 34,
+      "seconds": 0.9293085339995741,
+      "bytes": 2351431,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.1122738659996685,
-      "bytes": 11663489,
-      "pages": 30,
+      "seconds": 1.204088584998317,
+      "bytes": 7306502,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.1172626269999455,
-      "bytes": 11245903,
+      "seconds": 1.0363285590010491,
+      "bytes": 11683428,
       "pages": 36,
       "shards": 4
     },
     {
-      "seconds": 0.9253301299995655,
-      "bytes": 2328336,
-      "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.028129626000009,
-      "bytes": 6087526,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.0476808559997153,
+      "bytes": 995690,
+      "pages": 29,
+      "shards": 1
     },
     {
-      "seconds": 0.7597279170004185,
-      "bytes": 2156570,
+      "seconds": 1.5069925719999446,
+      "bytes": 1775692,
       "pages": 21,
       "shards": 2
     },
     {
-      "seconds": 0.9871070610006427,
-      "bytes": 12008396,
-      "pages": 45,
+      "seconds": 1.199250067000321,
+      "bytes": 13436756,
+      "pages": 47,
       "shards": 4
     },
     {
-      "seconds": 1.228597503999481,
-      "bytes": 13077636,
+      "seconds": 1.2015854260007472,
+      "bytes": 10309364,
       "pages": 34,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.1013024290004978,
-      "bytes": 1132593,
-      "pages": 33,
-      "shards": 1
+      "seconds": 0.9169029310014594,
+      "bytes": 7362656,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.9999866519992793,
-      "bytes": 6918002,
-      "pages": 32,
-      "shards": 3
+      "seconds": 0.9597948790014925,
+      "bytes": 2508350,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.2065209279990086,
-      "bytes": 11423370,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.1389318259989523,
+      "bytes": 6045358,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.0073676210013218,
-      "bytes": 932900,
+      "seconds": 1.1304748169986851,
+      "bytes": 12071070,
       "pages": 27,
-      "shards": 1
+      "shards": 4
     },
     {
-      "seconds": 1.0279631710000103,
-      "bytes": 7068347,
-      "pages": 35,
+      "seconds": 3.228665867998643,
+      "bytes": 7676896,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.8251235360003193,
-      "bytes": 2046300,
-      "pages": 19,
+      "seconds": 0.9868028299988509,
+      "bytes": 6101148,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.6571312910000415,
-      "bytes": 16761254,
-      "pages": 38,
-      "shards": 4
+      "seconds": 0.8392134009991423,
+      "bytes": 3718175,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 0.7977973630004271,
-      "bytes": 2044807,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.0338640699992538,
+      "bytes": 9654277,
+      "pages": 35,
+      "shards": 3
+    },
+    {
+      "seconds": 0.8135462400005053,
+      "bytes": 1110572,
+      "pages": 33,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3182481060011924,
+      "bytes": 8487979,
+      "pages": 27,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:809e3a6fcad791a5e8798c9628745256615bd83b94d0bf2a79256dfb46e2c777
-size 721783295
+oid sha256:9ee3cdbc8544ee947af4a7d860b9cd7d6a5386d7eb87fa5e677065b54efe108a
+size 721787967