Qwen90M step 425000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 420000,
+  "step": 425000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "cfce39253aa3ae4c325817f7ed5dea73b4ff79ab53ddb17407c5d79b88e8a80a",
+    "model.safetensors": "e6a8119feebeb9b4f48e3d9e2dc5415b553b928301f4bcae69173e0b92d62b00",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "9af8f113fcd330718ea1f4faa81d21b284817f06e6050316baaf832fa2400fa0",
+    "trainer.pt": "b280b0d64c4eb0d14a1021d0c9cdbedc4c0bf8eb245a63c69257180b96af6158",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18",
-    "shard_scheduler.json": "b660820a9781b36ec51f7d77448e3fea7f7e2d6ad47c4a9335f882c29c9b5655",
-    "coverage.json": "c74982ad4eb4d14007b9313e674f56558ace0621e80565c1e8d7fd9ccd3565be",
+    "shard_scheduler.json": "53c18fdc60e1935cc354da354050e76668d361282678ac8d9bee2b8f0811a722",
+    "coverage.json": "3b07084bb276dfdd5fdda54e8af8ef1840ed097f7c487c84187259f480a569a9",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 420000,
+  "step": 425000,
   "stats": {
-    "climbmix/blocks": 15120000,
-    "climbmix/targets": 25276172773,
-    "climbmix/input_tokens": 25277773647,
-    "ultra_qa/blocks": 5670000,
-    "ultra_qa/targets": 9249409599,
-    "ultra_qa/input_tokens": 9249427358,
-    "cortex/blocks": 5670000,
-    "cortex/targets": 215646829,
-    "cortex/input_tokens": 9013142666,
-    "ultra_style/blocks": 5670000,
-    "ultra_style/targets": 8968862759,
-    "ultra_style/input_tokens": 8969333094,
-    "openmath/blocks": 3780000,
-    "openmath/targets": 5841040378,
-    "openmath/input_tokens": 5842986747,
-    "rewrite6/blocks": 1890000,
-    "rewrite6/targets": 3098343100,
-    "rewrite6/input_tokens": 3101740258
+    "climbmix/blocks": 15300000,
+    "climbmix/targets": 25576792778,
+    "climbmix/input_tokens": 25578412852,
+    "ultra_qa/blocks": 5737500,
+    "ultra_qa/targets": 9359436204,
+    "ultra_qa/input_tokens": 9359454150,
+    "cortex/blocks": 5737500,
+    "cortex/targets": 218211819,
+    "cortex/input_tokens": 9120392161,
+    "ultra_style/blocks": 5737500,
+    "ultra_style/targets": 9075454296,
+    "ultra_style/input_tokens": 9075929780,
+    "openmath/blocks": 3825000,
+    "openmath/targets": 5910673230,
+    "openmath/input_tokens": 5912642746,
+    "rewrite6/blocks": 1912500,
+    "rewrite6/targets": 3135273649,
+    "rewrite6/input_tokens": 3138711114
   },
   "sources": {
     "climbmix": {
-      "page": 2070,
-      "row": 2319,
+      "page": 2095,
+      "row": 1309,
       "start": {
-        "file": 100,
+        "file": 101,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 17,
+          "row_group": 34,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 8481039
+      "documents": 8582429
     },
     "rewrite6": {
-      "page": 28,
-      "row": 741,
+      "page": 42,
+      "row": 888,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 114688
+          "row": 172032
         }
       },
       "epoch": 23,
-      "offset": 2048,
-      "documents": 4839721
+      "offset": 0,
+      "documents": 4897212
     },
     "openmath": {
-      "page": 1294,
-      "row": 601,
+      "page": 1309,
+      "row": 2292,
       "start": {
         "file": 12,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 60,
-          "row": 424
+          "row_group": 121,
+          "row": 864
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 5300825
+      "documents": 5363956
     },
     "ultra_qa": {
-      "page": 559,
-      "row": 2275,
+      "page": 566,
+      "row": 874,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 3,
-          "row": 17426
+          "row": 46098
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 2291939
+      "documents": 2319210
     },
     "ultra_style": {
-      "page": 854,
-      "row": 414,
+      "page": 864,
+      "row": 304,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 0,
-          "row": 73272
+          "row_group": 1,
+          "row": 29158
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3498398
+      "documents": 3539248
     },
     "cortex": {
-      "page": 109970,
-      "row": 5,
+      "page": 111278,
+      "row": 123,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 14076165
+      "documents": 14243707
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 420000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "cfce39253aa3ae4c325817f7ed5dea73b4ff79ab53ddb17407c5d79b88e8a80a", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "9af8f113fcd330718ea1f4faa81d21b284817f06e6050316baaf832fa2400fa0", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18", "shard_scheduler.json": "b660820a9781b36ec51f7d77448e3fea7f7e2d6ad47c4a9335f882c29c9b5655", "coverage.json": "c74982ad4eb4d14007b9313e674f56558ace0621e80565c1e8d7fd9ccd3565be", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000420000"}
+{"step": 425000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e6a8119feebeb9b4f48e3d9e2dc5415b553b928301f4bcae69173e0b92d62b00", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "b280b0d64c4eb0d14a1021d0c9cdbedc4c0bf8eb245a63c69257180b96af6158", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "a6ae17d5ca4d76a29c6e069dddb352125dfbf6f18e59e9553ab94c741a962b18", "shard_scheduler.json": "53c18fdc60e1935cc354da354050e76668d361282678ac8d9bee2b8f0811a722", "coverage.json": "3b07084bb276dfdd5fdda54e8af8ef1840ed097f7c487c84187259f480a569a9", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000425000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cfce39253aa3ae4c325817f7ed5dea73b4ff79ab53ddb17407c5d79b88e8a80a
+oid sha256:e6a8119feebeb9b4f48e3d9e2dc5415b553b928301f4bcae69173e0b92d62b00
 size 360622744

shard_scheduler.json

@@ -22,8 +22,8 @@
       "tokens_per_second": 598315.063954628,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 77052.14963302753,
-      "required_pages_per_second_upper_estimate": 0.005330646497231844
+      "required_token_positions_per_second": 76980.58360649772,
+      "required_pages_per_second_upper_estimate": 0.005325695393460453
     },
     {
       "source": "rewrite6",
@@ -36,8 +36,8 @@
       "tokens_per_second": 8533.957213384612,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 9631.51870412844,
-      "required_pages_per_second_upper_estimate": 0.004032515752436897
+      "required_token_positions_per_second": 9622.572950812215,
+      "required_pages_per_second_upper_estimate": 0.004028770352331965
     },
     {
       "source": "openmath",
@@ -50,8 +50,8 @@
       "tokens_per_second": 380144.35761751566,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 19263.03740825688,
-      "required_pages_per_second_upper_estimate": 0.004011982691681902
+      "required_token_positions_per_second": 19245.14590162443,
+      "required_pages_per_second_upper_estimate": 0.004008256362681121
     },
     {
       "source": "ultra_qa",
@@ -64,8 +64,8 @@
       "tokens_per_second": 43742.16556775435,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28894.556112385326,
-      "required_pages_per_second_upper_estimate": 0.0017321845195049147
+      "required_token_positions_per_second": 28867.718852436643,
+      "required_pages_per_second_upper_estimate": 0.0017305756667491143
     },
     {
       "source": "ultra_style",
@@ -78,8 +78,8 @@
       "tokens_per_second": 22018.86276523721,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28894.556112385326,
-      "required_pages_per_second_upper_estimate": 0.002246424644017299
+      "required_token_positions_per_second": 28867.718852436643,
+      "required_pages_per_second_upper_estimate": 0.00224433816510092
     },
     {
       "source": "cortex",
@@ -92,8 +92,8 @@
       "tokens_per_second": 910315.184299418,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 23,
-      "required_token_positions_per_second": 28894.556112385326,
-      "required_pages_per_second_upper_estimate": 0.3706283412524894
+      "required_token_positions_per_second": 28867.718852436643,
+      "required_pages_per_second_upper_estimate": 0.3702841016974724
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 704566,
-    "producer_batch_seconds": 0.09326356839858363,
-    "producer_p95_seconds": 0.27388932800022303,
-    "gpu_update_seconds": 0.956858443938822,
-    "producer_batches_per_second": 10.722300434895077,
-    "gpu_batches_per_second": 1.0450866649444923,
-    "producer_headroom": 10.259723709578259,
+    "producer_batch_seconds": 0.4470424027734907,
+    "producer_p95_seconds": 0.30381212399879587,
+    "gpu_update_seconds": 0.9577480001564553,
+    "producer_batches_per_second": 2.23692426891926,
+    "gpu_batches_per_second": 1.0441159885863949,
+    "producer_headroom": 2.142409745058862,
     "sustainable": true,
-    "buffer_seconds": 60.28208196814579,
+    "buffer_seconds": 60.338124009856685,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 82,
+    "upload_pages_per_commit": 77,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 376352.5842703375,
-    "upload_bytes_per_second": 5020670.556623916,
+    "generation_bytes_per_second": 399755.21465372964,
+    "upload_bytes_per_second": 5963845.868077777,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 80.75650354464143,
+    "estimated_full_shards_per_hour": 85.77816323956411,
     "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": 1.2489348130002327,
-      "bytes": 8291174,
-      "pages": 32,
-      "shards": 3
-    },
-    {
-      "seconds": 1.122991837999507,
-      "bytes": 10369739,
-      "pages": 39,
+      "seconds": 1.0381844509975053,
+      "bytes": 7741032,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.367829029000859,
-      "bytes": 6101351,
-      "pages": 37,
+      "seconds": 1.5200233909999952,
+      "bytes": 6160062,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9580737389987917,
-      "bytes": 2350217,
-      "pages": 28,
+      "seconds": 1.1740434719977202,
+      "bytes": 8468925,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.1762084339989087,
-      "bytes": 10342096,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.0248841799984802,
+      "bytes": 5963815,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 1.164775548999387,
-      "bytes": 5765223,
-      "pages": 29,
+      "seconds": 1.231629100999271,
+      "bytes": 4969209,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9962533819998498,
-      "bytes": 8524201,
-      "pages": 41,
+      "seconds": 1.249853739001992,
+      "bytes": 7694709,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.0446712169996317,
-      "bytes": 5880079,
+      "seconds": 1.2034739760019875,
+      "bytes": 5697150,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.5804328610010998,
-      "bytes": 12088466,
-      "pages": 42,
-      "shards": 4
+      "seconds": 0.9529857010020351,
+      "bytes": 6624046,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 0.7921914839989768,
-      "bytes": 902002,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.1070565920017543,
+      "bytes": 11831537,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.0630905629986955,
-      "bytes": 6092917,
+      "seconds": 1.0998464429976593,
+      "bytes": 6314908,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9817484880004486,
-      "bytes": 2589304,
-      "pages": 36,
+      "seconds": 1.0093370079994202,
+      "bytes": 2319025,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2340881659984007,
-      "bytes": 16290946,
-      "pages": 38,
-      "shards": 4
-    },
-    {
-      "seconds": 1.10198693500206,
-      "bytes": 7334244,
-      "pages": 30,
+      "seconds": 1.116415386000881,
+      "bytes": 10937725,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 0.7830869729987171,
-      "bytes": 1161196,
-      "pages": 35,
+      "seconds": 0.8731399669995881,
+      "bytes": 778500,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 1.0245182449980348,
-      "bytes": 6075842,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.033372367000993,
+      "bytes": 7683491,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.3879416819981998,
-      "bytes": 11260844,
+      "seconds": 1.4393485709988454,
+      "bytes": 11721944,
       "pages": 36,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.0456256510005915,
-      "bytes": 862691,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.162564084999758,
+      "bytes": 5892604,
+      "pages": 23,
+      "shards": 3
     },
     {
-      "seconds": 2.013755919000687,
-      "bytes": 11740969,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.082973058000789,
+      "bytes": 6647343,
+      "pages": 45,
+      "shards": 2
     },
     {
-      "seconds": 0.9664811299990106,
-      "bytes": 2719675,
+      "seconds": 1.1072141639997426,
+      "bytes": 6205762,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2011403209980926,
-      "bytes": 11095830,
-      "pages": 34,
+      "seconds": 0.9200315849993785,
+      "bytes": 2298271,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.584766658001172,
+      "bytes": 10894182,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0495139129998279,
-      "bytes": 7453354,
+      "seconds": 1.1065202139980101,
+      "bytes": 8432247,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.6562018110016652,
-      "bytes": 1121250,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.2434194350025791,
+      "bytes": 6179718,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.9384285160012951,
-      "bytes": 5824612,
-      "pages": 28,
+      "seconds": 1.03145402500013,
+      "bytes": 6170934,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.8717172680007934,
-      "bytes": 283568,
-      "pages": 8,
+      "seconds": 0.9970344739995198,
+      "bytes": 6228444,
+      "pages": 33,
+      "shards": 3
+    },
+    {
+      "seconds": 0.8807951420021709,
+      "bytes": 343455,
+      "pages": 9,
       "shards": 1
     },
     {
-      "seconds": 14.44776459899731,
-      "bytes": 11619395,
-      "pages": 28,
-      "shards": 4
+      "seconds": 1.1079528180016496,
+      "bytes": 5894448,
+      "pages": 27,
+      "shards": 3
     },
     {
-      "seconds": 1.2608952039990982,
-      "bytes": 6128515,
-      "pages": 36,
+      "seconds": 1.0535939100009273,
+      "bytes": 7640568,
+      "pages": 40,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9312373259999731,
+      "bytes": 6794347,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 0.9641569820014411,
-      "bytes": 6035455,
+      "seconds": 1.2384252730007574,
+      "bytes": 10759915,
       "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 0.8831225600006292,
+      "bytes": 2598975,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.0038025159992685,
-      "bytes": 2414763,
+      "seconds": 1.0244675720023224,
+      "bytes": 5791149,
       "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.1136625049985014,
-      "bytes": 10469225,
+      "seconds": 1.2756652919997578,
+      "bytes": 7797831,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9355267249993631,
-      "bytes": 7522148,
-      "pages": 39,
+      "seconds": 0.9431873399989854,
+      "bytes": 5431625,
+      "pages": 32,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3104591960000107,
+      "bytes": 11823249,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.2877170510000724,
-      "bytes": 6767236,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.1533994220008026,
+      "bytes": 7669461,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.0314974919965607,
-      "bytes": 5945434,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.8288239209978201,
+      "bytes": 961455,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.1649709810008062,
-      "bytes": 7159686,
-      "pages": 37,
+      "seconds": 1.3137797929994122,
+      "bytes": 10288628,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.206743112001277,
-      "bytes": 6257240,
-      "pages": 36,
+      "seconds": 1.1022804700005508,
+      "bytes": 2387118,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.0216239910005243,
-      "bytes": 7839111,
-      "pages": 37,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0000162349970196,
-      "bytes": 885768,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.1524969450001663,
+      "bytes": 6155701,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.2385705720007536,
-      "bytes": 15798086,
+      "seconds": 1.4659404199992423,
+      "bytes": 13390262,
       "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 0.9273680320002313,
-      "bytes": 2684614,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.003430278000451,
-      "bytes": 5945767,
-      "pages": 28,
+      "seconds": 0.8803229269979056,
+      "bytes": 4831371,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.8994475729996338,
-      "bytes": 6104947,
+      "seconds": 1.1803598539991071,
+      "bytes": 6473769,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0381844509975053,
-      "bytes": 7741032,
+      "seconds": 1.0211126419999346,
+      "bytes": 2557759,
       "pages": 38,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.5200233909999952,
-      "bytes": 6160062,
+      "seconds": 1.475009352001507,
+      "bytes": 6011684,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1740434719977202,
-      "bytes": 8468925,
-      "pages": 41,
+      "seconds": 1.40803722399869,
+      "bytes": 10317954,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.0248841799984802,
-      "bytes": 5963815,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.058006204002595,
+      "bytes": 8327609,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.231629100999271,
-      "bytes": 4969209,
+      "seconds": 1.1640828560011869,
+      "bytes": 6135465,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.249853739001992,
-      "bytes": 7694709,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2034739760019875,
-      "bytes": 5697150,
+      "seconds": 0.8637312579994614,
+      "bytes": 2324435,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9529857010020351,
-      "bytes": 6624046,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1070565920017543,
-      "bytes": 11831537,
-      "pages": 32,
+      "seconds": 1.1669898459986143,
+      "bytes": 10131671,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 1.0998464429976593,
-      "bytes": 6314908,
-      "pages": 36,
+      "seconds": 1.0318543029970897,
+      "bytes": 5757385,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0093370079994202,
-      "bytes": 2319025,
-      "pages": 29,
+      "seconds": 0.9449253849998058,
+      "bytes": 2649946,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.116415386000881,
-      "bytes": 10937725,
-      "pages": 41,
+      "seconds": 1.2970311710014357,
+      "bytes": 11562430,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.8731399669995881,
-      "bytes": 778500,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.4141549599989958,
+      "bytes": 12336426,
+      "pages": 42,
+      "shards": 4
     },
     {
-      "seconds": 1.033372367000993,
-      "bytes": 7683491,
-      "pages": 38,
-      "shards": 3
+      "seconds": 0.8797184569993988,
+      "bytes": 870056,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.4393485709988454,
-      "bytes": 11721944,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.3180106779982452,
+      "bytes": 6077154,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.162564084999758,
-      "bytes": 5892604,
-      "pages": 23,
-      "shards": 3
+      "seconds": 0.8500619599981292,
+      "bytes": 2427201,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 1.082973058000789,
-      "bytes": 6647343,
-      "pages": 45,
-      "shards": 2
+      "seconds": 1.7313189029991918,
+      "bytes": 10259217,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.1072141639997426,
-      "bytes": 6205762,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.5030511410004692,
+      "bytes": 13228107,
+      "pages": 34,
+      "shards": 4
     },
     {
-      "seconds": 0.9200315849993785,
-      "bytes": 2298271,
-      "pages": 29,
-      "shards": 2
+      "seconds": 0.8756032579985913,
+      "bytes": 1232264,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.584766658001172,
-      "bytes": 10894182,
-      "pages": 37,
+      "seconds": 1.2522269889996096,
+      "bytes": 10696457,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.1065202139980101,
-      "bytes": 8432247,
-      "pages": 39,
+      "seconds": 1.1242637430004834,
+      "bytes": 7345030,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.2434194350025791,
-      "bytes": 6179718,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.8663610330004303,
+      "bytes": 1077995,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.03145402500013,
-      "bytes": 6170934,
-      "pages": 37,
+      "seconds": 0.9089211409991549,
+      "bytes": 5857565,
+      "pages": 34,
       "shards": 2
+    },
+    {
+      "seconds": 1.1926017650002905,
+      "bytes": 13608076,
+      "pages": 35,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9af8f113fcd330718ea1f4faa81d21b284817f06e6050316baaf832fa2400fa0
-size 721762431
+oid sha256:b280b0d64c4eb0d14a1021d0c9cdbedc4c0bf8eb245a63c69257180b96af6158
+size 721780415