Qwen90M step 820000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 815000,
+  "step": 820000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "fdf8df44c98676caf03f6204fff629f95101410723ef61ce23662a551a8f427b",
+    "model.safetensors": "be65b8dd85d32e3c046a0a29d18dd9c9636bd3518c6180704165ccc3b345f97a",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "125e3addd8b07f3024e8ed1e3c3e2f705e99c3cfcfa6b9f1547db31a2081c525",
+    "trainer.pt": "72a6fa391959ac41be57caac41e5d9e33fa242f668e7e959e3b49726861e8403",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7",
-    "shard_scheduler.json": "d8c00298782e3f8a28914751297b8d31ce17e96db9eaa5de1147c633462ccb8c",
-    "coverage.json": "1e90d10c39719e43af2b2ea2d77129d1f5aa16ce1b06969bdac4ca8d9ac3d0a6",
+    "shard_scheduler.json": "cca80237086820dcf5c494e00db56e92fb63f5bff088142a981f0d3db418bb87",
+    "coverage.json": "c13407e1cced45349ea4cc9f52ed38e019bd1fa9d3e900afc3652ad06d436a2c",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 815000,
+  "step": 820000,
   "stats": {
-    "climbmix/blocks": 29340000,
-    "climbmix/targets": 49047052329,
-    "climbmix/input_tokens": 49050160293,
-    "ultra_qa/blocks": 11002500,
-    "ultra_qa/targets": 17949369763,
-    "ultra_qa/input_tokens": 17949404365,
-    "cortex/blocks": 11002500,
-    "cortex/targets": 418371734,
-    "cortex/input_tokens": 17489228968,
-    "ultra_style/blocks": 11002500,
-    "ultra_style/targets": 17402557074,
-    "ultra_style/input_tokens": 17403469802,
-    "openmath/blocks": 7335000,
-    "openmath/targets": 11333868401,
-    "openmath/input_tokens": 11337644586,
-    "rewrite6/blocks": 3667500,
-    "rewrite6/targets": 6012349385,
-    "rewrite6/input_tokens": 6018941143
+    "climbmix/blocks": 29520000,
+    "climbmix/targets": 49348064056,
+    "climbmix/input_tokens": 49351191260,
+    "ultra_qa/blocks": 11070000,
+    "ultra_qa/targets": 18059605638,
+    "ultra_qa/input_tokens": 18059640442,
+    "cortex/blocks": 11070000,
+    "cortex/targets": 420935370,
+    "cortex/input_tokens": 17596608233,
+    "ultra_style/blocks": 11070000,
+    "ultra_style/targets": 17509519663,
+    "ultra_style/input_tokens": 17510438016,
+    "openmath/blocks": 7380000,
+    "openmath/targets": 11403359242,
+    "openmath/input_tokens": 11407158801,
+    "rewrite6/blocks": 3690000,
+    "rewrite6/targets": 6049287294,
+    "rewrite6/input_tokens": 6055919347
   },
   "sources": {
     "climbmix": {
-      "page": 4018,
-      "row": 620,
+      "page": 4042,
+      "row": 3493,
       "start": {
-        "file": 194,
+        "file": 195,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 41,
+          "row_group": 54,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 8192,
-      "documents": 16458348
+      "offset": 2048,
+      "documents": 16559525
     },
     "rewrite6": {
-      "page": 36,
-      "row": 343,
+      "page": 50,
+      "row": 410,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 147456
+          "row": 204800
         }
       },
       "epoch": 45,
-      "offset": 2048,
-      "documents": 9390979
+      "offset": 0,
+      "documents": 9448390
     },
     "openmath": {
-      "page": 2511,
-      "row": 1078,
+      "page": 2526,
+      "row": 3013,
       "start": {
         "file": 23,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 242,
-          "row": 106
+          "row_group": 303,
+          "row": 546
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 10286134
+      "documents": 10349509
     },
     "ultra_qa": {
-      "page": 1085,
-      "row": 3490,
+      "page": 1092,
+      "row": 2051,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 28928
+          "row": 57600
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 4447650
+      "offset": 2048,
+      "documents": 4474883
     },
     "ultra_style": {
-      "page": 1657,
-      "row": 1167,
+      "page": 1667,
+      "row": 1796,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 27375
+          "row": 68335
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 6788239
+      "offset": 2048,
+      "documents": 6829828
     },
     "cortex": {
-      "page": 213380,
-      "row": 6,
+      "page": 214689,
+      "row": 86,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 27312646
+      "documents": 27480278
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 815000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "fdf8df44c98676caf03f6204fff629f95101410723ef61ce23662a551a8f427b", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "125e3addd8b07f3024e8ed1e3c3e2f705e99c3cfcfa6b9f1547db31a2081c525", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "d8c00298782e3f8a28914751297b8d31ce17e96db9eaa5de1147c633462ccb8c", "coverage.json": "1e90d10c39719e43af2b2ea2d77129d1f5aa16ce1b06969bdac4ca8d9ac3d0a6", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000815000"}
+{"step": 820000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "be65b8dd85d32e3c046a0a29d18dd9c9636bd3518c6180704165ccc3b345f97a", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "72a6fa391959ac41be57caac41e5d9e33fa242f668e7e959e3b49726861e8403", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "cca80237086820dcf5c494e00db56e92fb63f5bff088142a981f0d3db418bb87", "coverage.json": "c13407e1cced45349ea4cc9f52ed38e019bd1fa9d3e900afc3652ad06d436a2c", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000820000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fdf8df44c98676caf03f6204fff629f95101410723ef61ce23662a551a8f427b
+oid sha256:be65b8dd85d32e3c046a0a29d18dd9c9636bd3518c6180704165ccc3b345f97a
 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": 76422.60943566434,
-      "required_pages_per_second_upper_estimate": 0.005287093445643752
+      "required_token_positions_per_second": 76942.25922787549,
+      "required_pages_per_second_upper_estimate": 0.005323044024022557
     },
     {
       "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": 9552.826179458043,
-      "required_pages_per_second_upper_estimate": 0.003999568835644181
+      "required_token_positions_per_second": 9617.782403484436,
+      "required_pages_per_second_upper_estimate": 0.004026764650203828
     },
     {
       "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": 19105.652358916086,
-      "required_pages_per_second_upper_estimate": 0.003979203536427075
+      "required_token_positions_per_second": 19235.564806968872,
+      "required_pages_per_second_upper_estimate": 0.004006260873334826
     },
     {
       "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": 28658.47853837413,
-      "required_pages_per_second_upper_estimate": 0.001718032029412527
+      "required_token_positions_per_second": 28853.347210453307,
+      "required_pages_per_second_upper_estimate": 0.00172971410875644
     },
     {
       "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": 28658.47853837413,
-      "required_pages_per_second_upper_estimate": 0.0022280706510334355
+      "required_token_positions_per_second": 28853.347210453307,
+      "required_pages_per_second_upper_estimate": 0.0022432208331508883
     },
     {
       "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": 28658.47853837413,
-      "required_pages_per_second_upper_estimate": 0.3676001916134238
+      "required_token_positions_per_second": 28853.347210453307,
+      "required_pages_per_second_upper_estimate": 0.37009975770517706
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,409 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.2541760841795906,
-    "producer_p95_seconds": 0.2388652240006195,
-    "gpu_update_seconds": 0.9647406774570715,
-    "producer_batches_per_second": 3.9342804545428445,
-    "gpu_batches_per_second": 1.0365479795418884,
-    "producer_headroom": 3.795560391021779,
+    "producer_batch_seconds": 0.08518336581272479,
+    "producer_p95_seconds": 0.2713440530023945,
+    "gpu_update_seconds": 0.958225047455963,
+    "producer_batches_per_second": 11.739381162731878,
+    "gpu_batches_per_second": 1.0435961809336411,
+    "producer_headroom": 11.248969071762392,
     "sustainable": true,
-    "buffer_seconds": 60.7786626797955,
+    "buffer_seconds": 60.36817798972567,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 82,
+    "upload_pages_per_commit": 78,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 448277.4506157306,
-    "upload_bytes_per_second": 3886909.442531961,
+    "generation_bytes_per_second": 237498.7448770881,
+    "upload_bytes_per_second": 3973102.9977444257,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 96.18990553716601,
+    "estimated_full_shards_per_hour": 50.96170196279986,
     "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": 1.4946416839993617,
-      "bytes": 5892242,
-      "pages": 27,
-      "shards": 2
-    },
-    {
-      "seconds": 1.5706974580007227,
-      "bytes": 6564398,
-      "pages": 38,
+      "seconds": 1.6904037299991614,
+      "bytes": 12097177,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.5560058639985073,
-      "bytes": 6086581,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.2079874939990987,
+      "bytes": 1225243,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.8728917230000661,
-      "bytes": 6075781,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.7503577689985832,
+      "bytes": 11722749,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.5521995129984134,
-      "bytes": 2545582,
-      "pages": 34,
+      "seconds": 1.717468271999678,
+      "bytes": 6201366,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.8264212399990356,
-      "bytes": 16670948,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.4180366079999658,
+      "bytes": 818969,
+      "pages": 25,
+      "shards": 1
     },
     {
-      "seconds": 1.714815281000483,
-      "bytes": 7600635,
+      "seconds": 1.935665021999739,
+      "bytes": 7829235,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.3434111860005942,
-      "bytes": 1056341,
-      "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 1.5977993900014553,
-      "bytes": 5977865,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.8666853409995383,
-      "bytes": 6845301,
-      "pages": 39,
+      "seconds": 1.7852341269990575,
+      "bytes": 11660999,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.52379110199945,
-      "bytes": 6381300,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.8856627909990493,
+      "bytes": 7118698,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.4397832380000182,
-      "bytes": 6693273,
-      "pages": 27,
+      "seconds": 1.6447518009990745,
+      "bytes": 6129393,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.380750825999712,
-      "bytes": 7212989,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.743864690999544,
+      "bytes": 1054003,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.7552309799993964,
-      "bytes": 9939009,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.5393589430004795,
+      "bytes": 11424070,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.423224260999632,
-      "bytes": 2660337,
-      "pages": 36,
+      "seconds": 1.9194490070003667,
+      "bytes": 6288888,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.6043403389994637,
-      "bytes": 6329322,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.4686280049991183,
+      "bytes": 8470485,
+      "pages": 37,
+      "shards": 4
     },
     {
-      "seconds": 1.7717889699997613,
-      "bytes": 5364175,
-      "pages": 27,
+      "seconds": 1.9930366190001223,
+      "bytes": 5866239,
+      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 1.5835829299994657,
-      "bytes": 13733184,
-      "pages": 40,
-      "shards": 4
+      "seconds": 2.011558195001271,
+      "bytes": 5842602,
+      "pages": 41,
+      "shards": 2
     },
     {
-      "seconds": 1.5670746849991701,
-      "bytes": 6380319,
+      "seconds": 1.6693870859999151,
+      "bytes": 7612861,
       "pages": 29,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.1554255430000921,
-      "bytes": 1163617,
-      "pages": 34,
+      "seconds": 1.3870315900003334,
+      "bytes": 1261477,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 2.120087691999288,
-      "bytes": 11751236,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.6420892379992438,
+      "bytes": 7607765,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.3820675099996151,
-      "bytes": 1056686,
-      "pages": 32,
-      "shards": 1
+      "seconds": 1.5253431410001212,
+      "bytes": 5688723,
+      "pages": 19,
+      "shards": 2
     },
     {
-      "seconds": 1.7487962389986933,
-      "bytes": 7557748,
-      "pages": 38,
-      "shards": 4
+      "seconds": 1.593651556999248,
+      "bytes": 11614889,
+      "pages": 46,
+      "shards": 3
     },
     {
-      "seconds": 1.7842456250000396,
-      "bytes": 12131753,
-      "pages": 31,
+      "seconds": 1.818591990000641,
+      "bytes": 7461958,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.6632540899991,
-      "bytes": 5726975,
-      "pages": 37,
+      "seconds": 1.4790617950002343,
+      "bytes": 1048131,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3689376410002296,
+      "bytes": 5834070,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.7110617899998033,
-      "bytes": 7430334,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.2757619459989655,
+      "bytes": 12484592,
+      "pages": 45,
+      "shards": 4
     },
     {
-      "seconds": 1.6944532489997073,
-      "bytes": 1160058,
-      "pages": 35,
+      "seconds": 1.4620103640008892,
+      "bytes": 316633,
+      "pages": 8,
       "shards": 1
     },
     {
-      "seconds": 2.4445176089993765,
-      "bytes": 7341476,
-      "pages": 34,
+      "seconds": 1.8397933680007554,
+      "bytes": 6758434,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.9822834630012949,
+      "bytes": 7671850,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.587468388999696,
-      "bytes": 5869829,
-      "pages": 27,
+      "seconds": 1.43599132899908,
+      "bytes": 5917528,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.9299586070010264,
-      "bytes": 10891963,
+      "seconds": 1.7868269310019969,
+      "bytes": 5421842,
       "pages": 40,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.62565908400029,
-      "bytes": 7724112,
-      "pages": 39,
+      "seconds": 1.8639257160029956,
+      "bytes": 7235620,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.4196104830007243,
-      "bytes": 988295,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.8414086459997634,
+      "bytes": 5860537,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 2.0304724819998228,
-      "bytes": 10476032,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.8819817990006413,
+      "bytes": 6541422,
+      "pages": 23,
+      "shards": 2
     },
     {
-      "seconds": 1.5808353909997095,
-      "bytes": 7239609,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.735375592001219,
+      "bytes": 12429197,
+      "pages": 48,
+      "shards": 4
     },
     {
-      "seconds": 1.859183167000083,
-      "bytes": 1137645,
-      "pages": 35,
+      "seconds": 1.311772810000548,
+      "bytes": 931562,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 4.019235966999986,
-      "bytes": 13232606,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.873525888997392,
+      "bytes": 7538765,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.720672621000631,
-      "bytes": 611156,
-      "pages": 21,
-      "shards": 1
+      "seconds": 1.4924008759990102,
+      "bytes": 6022523,
+      "pages": 32,
+      "shards": 2
     },
     {
-      "seconds": 1.6194256800008588,
-      "bytes": 9820718,
-      "pages": 47,
+      "seconds": 1.568882242001564,
+      "bytes": 7013425,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.6360844549999456,
-      "bytes": 7454978,
-      "pages": 29,
+      "seconds": 1.719156404997193,
+      "bytes": 11941861,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.568186391999916,
-      "bytes": 1164036,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.5756286089999776,
+      "bytes": 6163874,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 2.5616059220010357,
-      "bytes": 11807815,
-      "pages": 42,
-      "shards": 4
+      "seconds": 1.85335265699905,
+      "bytes": 5996093,
+      "pages": 28,
+      "shards": 3
     },
     {
-      "seconds": 1.6904037299991614,
-      "bytes": 12097177,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.6912986729985278,
+      "bytes": 6223259,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.2079874939990987,
-      "bytes": 1225243,
-      "pages": 35,
+      "seconds": 1.3231591360017774,
+      "bytes": 1106930,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.7503577689985832,
-      "bytes": 11722749,
-      "pages": 33,
-      "shards": 4
+      "seconds": 2.400186033999489,
+      "bytes": 7382284,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.717468271999678,
-      "bytes": 6201366,
-      "pages": 37,
-      "shards": 2
+      "seconds": 2.1097241989991744,
+      "bytes": 16842562,
+      "pages": 44,
+      "shards": 5
     },
     {
-      "seconds": 1.4180366079999658,
-      "bytes": 818969,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.6551350050031033,
+      "bytes": 2311821,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.935665021999739,
-      "bytes": 7829235,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.769397926000238,
+      "bytes": 6044362,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.7852341269990575,
-      "bytes": 11660999,
+      "seconds": 1.4953640580024512,
+      "bytes": 6379384,
       "pages": 37,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.8856627909990493,
-      "bytes": 7118698,
-      "pages": 30,
+      "seconds": 1.7474238050017448,
+      "bytes": 6290249,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.6447518009990745,
-      "bytes": 6129393,
-      "pages": 37,
+      "seconds": 1.5282406320002337,
+      "bytes": 5939964,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.743864690999544,
-      "bytes": 1054003,
-      "pages": 35,
+      "seconds": 1.2798854280008527,
+      "bytes": 545338,
+      "pages": 18,
       "shards": 1
     },
     {
-      "seconds": 1.5393589430004795,
-      "bytes": 11424070,
-      "pages": 35,
+      "seconds": 1.8976086379989283,
+      "bytes": 13629725,
+      "pages": 49,
       "shards": 4
     },
     {
-      "seconds": 1.9194490070003667,
-      "bytes": 6288888,
-      "pages": 35,
+      "seconds": 1.6741066590002447,
+      "bytes": 5692006,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.4686280049991183,
-      "bytes": 8470485,
+      "seconds": 1.5403353149995382,
+      "bytes": 7551970,
       "pages": 37,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.9930366190001223,
-      "bytes": 5866239,
-      "pages": 26,
+      "seconds": 1.5439765690025524,
+      "bytes": 6266973,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 2.011558195001271,
-      "bytes": 5842602,
-      "pages": 41,
-      "shards": 2
+      "seconds": 1.459664652000356,
+      "bytes": 1024497,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 1.6693870859999151,
-      "bytes": 7612861,
-      "pages": 29,
-      "shards": 3
+      "seconds": 1.8518995710001036,
+      "bytes": 10973445,
+      "pages": 22,
+      "shards": 4
     },
     {
-      "seconds": 1.3870315900003334,
-      "bytes": 1261477,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.523874344999058,
+      "bytes": 6793396,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.6420892379992438,
-      "bytes": 7607765,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.551560215997597,
+      "bytes": 6347247,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.5253431410001212,
-      "bytes": 5688723,
-      "pages": 19,
+      "seconds": 1.3656726940025692,
+      "bytes": 2327821,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.593651556999248,
-      "bytes": 11614889,
-      "pages": 46,
+      "seconds": 1.6191135740009486,
+      "bytes": 10591273,
+      "pages": 42,
       "shards": 3
     },
     {
-      "seconds": 1.818591990000641,
-      "bytes": 7461958,
-      "pages": 30,
+      "seconds": 1.5766091279983812,
+      "bytes": 7345474,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.4790617950002343,
-      "bytes": 1048131,
-      "pages": 34,
+      "seconds": 1.195227366002655,
+      "bytes": 1189304,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.3689376410002296,
-      "bytes": 5834070,
-      "pages": 28,
+      "seconds": 2.000728038001398,
+      "bytes": 11892571,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.4507886779974797,
+      "bytes": 2119215,
+      "pages": 21,
       "shards": 2
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:125e3addd8b07f3024e8ed1e3c3e2f705e99c3cfcfa6b9f1547db31a2081c525
-size 721763839
+oid sha256:72a6fa391959ac41be57caac41e5d9e33fa242f668e7e959e3b49726861e8403
+size 721780159