Qwen90M step 840000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 835000,
+  "step": 840000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "434bc5e66416d00f38daf4d1c20bd1d7d620b0fc06f97108edac543c845dc140",
+    "model.safetensors": "3e144501faba5106e3f43a2e36d221a8630b9cdfb6837a8f976f58b534e86cfa",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "8fc02097e06a5d8dde4c92534118b36500d860341974727dbdf1235db028d4da",
+    "trainer.pt": "c1114d95865c877e56d4ca2db61e9232398961a900ec723d9fe48b310b10c038",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7",
-    "shard_scheduler.json": "15ec84a118722aa0da30b1eedddde0fa6f0d20effd89f99c89d67b1b955f9b9e",
-    "coverage.json": "2268abb420311a79a47ad9499884d787f6b2b1069f26033253f7830b5909fc05",
+    "shard_scheduler.json": "5d089facc2fe4bc290dfa5a25481443cf9a74f9c2860a1b57795e6bdc0906c9e",
+    "coverage.json": "abfd3dc053aa9107e89eeb9bfe2735dc1cd78f0a1640546271b969e4322c5588",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 835000,
+  "step": 840000,
   "stats": {
-    "climbmix/blocks": 30060000,
-    "climbmix/targets": 50250362065,
-    "climbmix/input_tokens": 50253546093,
-    "ultra_qa/blocks": 11272500,
-    "ultra_qa/targets": 18390084306,
-    "ultra_qa/input_tokens": 18390119687,
-    "cortex/blocks": 11272500,
-    "cortex/targets": 428641870,
-    "cortex/input_tokens": 17918505142,
-    "ultra_style/blocks": 11272500,
-    "ultra_style/targets": 17829475019,
-    "ultra_style/input_tokens": 17830410407,
-    "openmath/blocks": 7515000,
-    "openmath/targets": 11612035358,
-    "openmath/input_tokens": 11615904568,
-    "rewrite6/blocks": 3757500,
-    "rewrite6/targets": 6159911854,
-    "rewrite6/input_tokens": 6166665342
+    "climbmix/blocks": 30240000,
+    "climbmix/targets": 50551290330,
+    "climbmix/input_tokens": 50554493320,
+    "ultra_qa/blocks": 11340000,
+    "ultra_qa/targets": 18500378600,
+    "ultra_qa/input_tokens": 18500414184,
+    "cortex/blocks": 11340000,
+    "cortex/targets": 431213356,
+    "cortex/input_tokens": 18025811942,
+    "ultra_style/blocks": 11340000,
+    "ultra_style/targets": 17936310265,
+    "ultra_style/input_tokens": 17937251230,
+    "openmath/blocks": 7560000,
+    "openmath/targets": 11681504015,
+    "openmath/input_tokens": 11685396540,
+    "rewrite6/blocks": 3780000,
+    "rewrite6/targets": 6196813702,
+    "rewrite6/input_tokens": 6203607570
   },
   "sources": {
     "climbmix": {
-      "page": 4116,
-      "row": 3220,
+      "page": 4141,
+      "row": 1870,
       "start": {
-        "file": 199,
+        "file": 200,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 20,
+          "row_group": 37,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 16862356
+      "offset": 2048,
+      "documents": 16963406
     },
     "rewrite6": {
-      "page": 42,
-      "row": 790,
+      "page": 6,
+      "row": 281,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 172032
+          "row": 24576
         }
       },
-      "epoch": 46,
+      "epoch": 47,
       "offset": 0,
-      "documents": 9621406
+      "documents": 9678845
     },
     "openmath": {
-      "page": 2572,
-      "row": 4072,
+      "page": 2588,
+      "row": 1820,
       "start": {
         "file": 24,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 55,
-          "row": 313
+          "row_group": 120,
+          "row": 849
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 10538984
+      "documents": 10602268
     },
     "ultra_qa": {
-      "page": 1112,
-      "row": 1914,
+      "page": 1119,
+      "row": 459,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 6,
-          "row": 10507
+          "row": 39179
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 4556666
+      "offset": 0,
+      "documents": 4583883
     },
     "ultra_style": {
-      "page": 1698,
-      "row": 306,
+      "page": 1708,
+      "row": 803,
       "start": {
         "file": 10,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 20264
+          "row": 61224
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 6955314
+      "offset": 2048,
+      "documents": 6996771
     },
     "cortex": {
-      "page": 218617,
-      "row": 105,
+      "page": 219925,
+      "row": 120,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 27983081
+      "documents": 28150520
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 835000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "434bc5e66416d00f38daf4d1c20bd1d7d620b0fc06f97108edac543c845dc140", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "8fc02097e06a5d8dde4c92534118b36500d860341974727dbdf1235db028d4da", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "15ec84a118722aa0da30b1eedddde0fa6f0d20effd89f99c89d67b1b955f9b9e", "coverage.json": "2268abb420311a79a47ad9499884d787f6b2b1069f26033253f7830b5909fc05", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000835000"}
+{"step": 840000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "3e144501faba5106e3f43a2e36d221a8630b9cdfb6837a8f976f58b534e86cfa", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "c1114d95865c877e56d4ca2db61e9232398961a900ec723d9fe48b310b10c038", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "5d089facc2fe4bc290dfa5a25481443cf9a74f9c2860a1b57795e6bdc0906c9e", "coverage.json": "abfd3dc053aa9107e89eeb9bfe2735dc1cd78f0a1640546271b969e4322c5588", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000840000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:434bc5e66416d00f38daf4d1c20bd1d7d620b0fc06f97108edac543c845dc140
+oid sha256:3e144501faba5106e3f43a2e36d221a8630b9cdfb6837a8f976f58b534e86cfa
 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": 76407.9111220439,
-      "required_pages_per_second_upper_estimate": 0.0052860765822027025
+      "required_token_positions_per_second": 77053.7277135633,
+      "required_pages_per_second_upper_estimate": 0.005330755672504961
     },
     {
       "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": 9550.988890255487,
-      "required_pages_per_second_upper_estimate": 0.003998799601022032
+      "required_token_positions_per_second": 9631.715964195413,
+      "required_pages_per_second_upper_estimate": 0.0040325983411076795
     },
     {
       "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": 19101.977780510973,
-      "required_pages_per_second_upper_estimate": 0.00397843821865044
+      "required_token_positions_per_second": 19263.431928390826,
+      "required_pages_per_second_upper_estimate": 0.004012064859821606
     },
     {
       "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": 28652.966670766462,
-      "required_pages_per_second_upper_estimate": 0.001717701601365582
+      "required_token_positions_per_second": 28895.147892586243,
+      "required_pages_per_second_upper_estimate": 0.0017322199958243636
     },
     {
       "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": 28652.966670766462,
-      "required_pages_per_second_upper_estimate": 0.0022276421275710828
+      "required_token_positions_per_second": 28895.147892586243,
+      "required_pages_per_second_upper_estimate": 0.0022464706523249544
     },
     {
       "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": 28652.966670766462,
-      "required_pages_per_second_upper_estimate": 0.3675294912939349
+      "required_token_positions_per_second": 28895.147892586243,
+      "required_pages_per_second_upper_estimate": 0.3706359319735027
     }
   ],
   "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": 720750,
-    "producer_batch_seconds": 0.10879093923438177,
-    "producer_p95_seconds": 0.32612680799502414,
-    "gpu_update_seconds": 0.964926261133309,
-    "producer_batches_per_second": 9.191941967203505,
-    "gpu_batches_per_second": 1.0363486209044583,
-    "producer_headroom": 8.86954619496803,
+    "producer_batch_seconds": 0.15130693903114434,
+    "producer_p95_seconds": 0.37083707300189417,
+    "gpu_update_seconds": 0.956838847227142,
+    "producer_batches_per_second": 6.609082216607161,
+    "gpu_batches_per_second": 1.0451080690316203,
+    "producer_headroom": 6.3238266093678,
     "sustainable": true,
-    "buffer_seconds": 60.79035445139847,
+    "buffer_seconds": 60.280847375309946,
     "buffer_drain_seconds": null,
-    "shard_target_mib": 32,
-    "upload_pages_per_commit": 165,
+    "shard_target_mib": 16,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 666855.3792614836,
-    "upload_bytes_per_second": 3927796.2312513823,
+    "generation_bytes_per_second": 369362.01326889056,
+    "upload_bytes_per_second": 4183866.9114862867,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 71.54582039539042,
+    "estimated_full_shards_per_hour": 79.25648973989523,
     "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": 1.5374675419989217,
-      "bytes": 6254376,
+      "seconds": 2.2660662760026753,
+      "bytes": 11276737,
       "pages": 36,
+      "shards": 4
+    },
+    {
+      "seconds": 1.8010207819970674,
+      "bytes": 5858104,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.5695442620017275,
-      "bytes": 11600439,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.3410764899963397,
+      "bytes": 1019830,
+      "pages": 31,
+      "shards": 1
     },
     {
-      "seconds": 1.7552413789999264,
-      "bytes": 6467958,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.8211984559966368,
+      "bytes": 13136869,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 1.6743759019991558,
-      "bytes": 6275115,
-      "pages": 37,
+      "seconds": 1.4420257410019985,
+      "bytes": 5442386,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 2.272818253000878,
-      "bytes": 7589463,
-      "pages": 39,
+      "seconds": 2.6930293979967246,
+      "bytes": 5991867,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.5132595840004797,
-      "bytes": 5701329,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.7729905640007928,
+      "bytes": 7619842,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.5889989900024375,
-      "bytes": 6430899,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.2987675089971162,
+      "bytes": 1223490,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.6179743659995438,
-      "bytes": 8234542,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.546362834997126,
+      "bytes": 11822887,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.6133322549976583,
-      "bytes": 6059251,
-      "pages": 37,
+      "seconds": 1.5667999549987144,
+      "bytes": 6923893,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.6795359120005742,
-      "bytes": 5921876,
-      "pages": 24,
+      "seconds": 1.4826895270016394,
+      "bytes": 6163509,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.7577662710027653,
-      "bytes": 6866333,
-      "pages": 46,
+      "seconds": 1.7263165549957193,
+      "bytes": 7465725,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.93685000899859,
-      "bytes": 6023833,
-      "pages": 33,
+      "seconds": 2.012805174999812,
+      "bytes": 4595806,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.8011142590003146,
-      "bytes": 7499768,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.6402012090002245,
-      "bytes": 10865294,
-      "pages": 22,
+      "seconds": 1.6076512380022905,
+      "bytes": 7510724,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.7843240430011065,
-      "bytes": 6397585,
-      "pages": 42,
+      "seconds": 1.5883049809999648,
+      "bytes": 5665128,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.4411041519997525,
-      "bytes": 2727332,
-      "pages": 38,
+      "seconds": 2.0745964650050155,
+      "bytes": 11457524,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.913498541998706,
-      "bytes": 6177731,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 2.387397704002069,
-      "bytes": 11768966,
+      "seconds": 1.6211695970050641,
+      "bytes": 7634724,
       "pages": 40,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.479657814001257,
-      "bytes": 836473,
-      "pages": 27,
+      "seconds": 1.4634708860030514,
+      "bytes": 987981,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.9911568649986293,
-      "bytes": 11961950,
-      "pages": 42,
+      "seconds": 1.6518487119974452,
+      "bytes": 7534957,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.958790611999575,
-      "bytes": 2439502,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.5817108970004483,
+      "bytes": 10450177,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.8466758049980854,
-      "bytes": 10680095,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.328966004999529,
+      "bytes": 1184587,
+      "pages": 36,
+      "shards": 1
     },
     {
-      "seconds": 1.7479712950007524,
-      "bytes": 6009903,
-      "pages": 28,
-      "shards": 2
+      "seconds": 2.0095185120007955,
+      "bytes": 12961722,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 2.2005035470028815,
-      "bytes": 2644353,
+      "seconds": 1.4560003849983332,
+      "bytes": 6473170,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.5131899659972987,
-      "bytes": 5828306,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.6064382549957372,
+      "bytes": 6604109,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.8049468739991426,
-      "bytes": 4697774,
-      "pages": 26,
+      "seconds": 1.684055683996121,
+      "bytes": 5176791,
+      "pages": 11,
       "shards": 2
     },
     {
-      "seconds": 2.496297884001251,
-      "bytes": 13693480,
-      "pages": 42,
-      "shards": 4
+      "seconds": 1.8713450549985282,
+      "bytes": 942860,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.8265930059969833,
-      "bytes": 5855746,
-      "pages": 36,
+      "seconds": 1.5147400999994716,
+      "bytes": 6664610,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3942693540011533,
-      "bytes": 2663289,
-      "pages": 34,
+      "seconds": 1.4935083919990575,
+      "bytes": 2459570,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.5531271910003852,
-      "bytes": 10487346,
-      "pages": 37,
-      "shards": 3
+      "seconds": 2.1903198990039527,
+      "bytes": 16888519,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.5084796250012005,
-      "bytes": 1063958,
-      "pages": 32,
-      "shards": 1
+      "seconds": 2.158359011998982,
+      "bytes": 7618380,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.9720517980022123,
-      "bytes": 7314841,
+      "seconds": 1.546953492994362,
+      "bytes": 982772,
       "pages": 27,
-      "shards": 3
+      "shards": 1
     },
     {
-      "seconds": 1.793492888999026,
-      "bytes": 11753421,
+      "seconds": 1.5179005070021958,
+      "bytes": 6221682,
       "pages": 38,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.5141362069989555,
-      "bytes": 6622032,
-      "pages": 34,
+      "seconds": 1.4849864080024417,
+      "bytes": 6331011,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.5094903019999037,
-      "bytes": 6206747,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.7072867230017437,
+      "bytes": 11821281,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.7292475489994104,
-      "bytes": 6419421,
-      "pages": 37,
+      "seconds": 1.688430515001528,
+      "bytes": 5778338,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.6914732460027153,
-      "bytes": 2366813,
-      "pages": 27,
+      "seconds": 1.6583789540018188,
+      "bytes": 2527741,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.8487111699978414,
-      "bytes": 10763047,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.605599512000481,
-      "bytes": 8445757,
+      "seconds": 1.606587074995332,
+      "bytes": 10744249,
       "pages": 39,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 1.6673135730015929,
-      "bytes": 6027005,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.0940176939984667,
+      "bytes": 7184499,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.4991452859976562,
-      "bytes": 1121499,
+      "seconds": 1.0425019949980197,
+      "bytes": 1083508,
       "pages": 36,
       "shards": 1
     },
     {
-      "seconds": 2.2660662760026753,
-      "bytes": 11276737,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.5655531079974025,
+      "bytes": 12067871,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.8010207819970674,
-      "bytes": 5858104,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.9511204809969058,
+      "bytes": 6917854,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.3410764899963397,
-      "bytes": 1019830,
-      "pages": 31,
-      "shards": 1
+      "seconds": 1.1665768739985651,
+      "bytes": 5996687,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.8211984559966368,
-      "bytes": 13136869,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.6431592789958813,
+      "bytes": 7717315,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.4420257410019985,
-      "bytes": 5442386,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.3486441710047075,
+      "bytes": 976930,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 2.6930293979967246,
-      "bytes": 5991867,
-      "pages": 29,
+      "seconds": 1.5532393190005678,
+      "bytes": 10410458,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.7729905640007928,
-      "bytes": 7619842,
-      "pages": 39,
+      "seconds": 1.5896227589983027,
+      "bytes": 8024141,
+      "pages": 26,
       "shards": 3
     },
     {
-      "seconds": 1.2987675089971162,
-      "bytes": 1223490,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.570418814997538,
+      "bytes": 6148151,
+      "pages": 41,
+      "shards": 2
     },
     {
-      "seconds": 1.546362834997126,
-      "bytes": 11822887,
-      "pages": 33,
+      "seconds": 1.876006619000691,
+      "bytes": 11510011,
+      "pages": 30,
       "shards": 4
     },
     {
-      "seconds": 1.5667999549987144,
-      "bytes": 6923893,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.1747664850045112,
+      "bytes": 1216277,
+      "pages": 36,
+      "shards": 1
     },
     {
-      "seconds": 1.4826895270016394,
-      "bytes": 6163509,
-      "pages": 32,
+      "seconds": 1.0712760539972805,
+      "bytes": 6301436,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.7263165549957193,
-      "bytes": 7465725,
-      "pages": 39,
+      "seconds": 1.2888189970035455,
+      "bytes": 7385771,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 2.012805174999812,
-      "bytes": 4595806,
-      "pages": 30,
+      "seconds": 1.5881777600006899,
+      "bytes": 5494719,
+      "pages": 25,
       "shards": 2
     },
     {
-      "seconds": 1.6076512380022905,
-      "bytes": 7510724,
-      "pages": 40,
+      "seconds": 1.394116926996503,
+      "bytes": 11997464,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.5883049809999648,
-      "bytes": 5665128,
-      "pages": 27,
+      "seconds": 1.7342578780007898,
+      "bytes": 2696654,
+      "pages": 39,
       "shards": 2
     },
     {
-      "seconds": 2.0745964650050155,
-      "bytes": 11457524,
+      "seconds": 1.611772108997684,
+      "bytes": 5903925,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.512940188003995,
+      "bytes": 7372289,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.6211695970050641,
-      "bytes": 7634724,
-      "pages": 40,
-      "shards": 3
+      "seconds": 0.9355933799961349,
+      "bytes": 5063118,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.4634708860030514,
-      "bytes": 987981,
+      "seconds": 1.774848651002685,
+      "bytes": 5953016,
       "pages": 29,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.6518487119974452,
-      "bytes": 7534957,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.877187440994021,
+      "bytes": 13513556,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.5817108970004483,
-      "bytes": 10450177,
-      "pages": 33,
+      "seconds": 1.8802343610004755,
+      "bytes": 5656226,
+      "pages": 32,
+      "shards": 2
+    },
+    {
+      "seconds": 1.6476056760002393,
+      "bytes": 7839870,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.328966004999529,
-      "bytes": 1184587,
-      "pages": 36,
+      "seconds": 1.3661578879982699,
+      "bytes": 1247022,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 2.0095185120007955,
-      "bytes": 12961722,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.4103239969990682,
+      "bytes": 5948663,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.4560003849983332,
-      "bytes": 6473170,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.2626624109980185,
+      "bytes": 7374458,
+      "pages": 32,
+      "shards": 3
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8fc02097e06a5d8dde4c92534118b36500d860341974727dbdf1235db028d4da
-size 721778239
+oid sha256:c1114d95865c877e56d4ca2db61e9232398961a900ec723d9fe48b310b10c038
+size 721765631