Qwen90M step 800000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 795000,
+  "step": 800000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "c733b97c332695bc7f7abc60464b770db4d4fcc6b88b47a5682c5c028b69b18b",
+    "model.safetensors": "8dbf79dba5ff97079bc140e423cf40b0ef487b5e04e4c30f7760173437d08164",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "023c990bb11b42ee0e1c6492a696588ada4b8af501b7eebe7875277af2fde1e4",
+    "trainer.pt": "84eb35d156f31624644b274e41f57e3691a9338e88ee83073b01c24b4744a8cf",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d",
-    "shard_scheduler.json": "94c6529555273c279838970e2265eb9fd5f7d996a5c9d1e791a026e9904a5f57",
-    "coverage.json": "b9b4d36d0e722bc228f69164009d751a999eaf64d6b1a370460af2d29f35b1cc",
+    "shard_scheduler.json": "cc275791320ae7c7d3b065796c547bbaa39e7bc94e15c07e5ba7950b8dffb861",
+    "coverage.json": "666363f885b45a74f2b6a41ece402cf5806e0f534f2f95e854c20588602a307a",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 795000,
+  "step": 800000,
   "stats": {
-    "climbmix/blocks": 28620000,
-    "climbmix/targets": 47843123314,
-    "climbmix/input_tokens": 47846155406,
-    "ultra_qa/blocks": 10732500,
-    "ultra_qa/targets": 17508983971,
-    "ultra_qa/input_tokens": 17509017700,
-    "cortex/blocks": 10732500,
-    "cortex/targets": 408110127,
-    "cortex/input_tokens": 17060141294,
-    "ultra_style/blocks": 10732500,
-    "ultra_style/targets": 16975769338,
-    "ultra_style/input_tokens": 16976659443,
-    "openmath/blocks": 7155000,
-    "openmath/targets": 11055702871,
-    "openmath/input_tokens": 11059386403,
-    "rewrite6/blocks": 3577500,
-    "rewrite6/targets": 5864763458,
-    "rewrite6/input_tokens": 5871193549
+    "climbmix/blocks": 28800000,
+    "climbmix/targets": 48143279504,
+    "climbmix/input_tokens": 48146330672,
+    "ultra_qa/blocks": 10800000,
+    "ultra_qa/targets": 17619033390,
+    "ultra_qa/input_tokens": 17619067325,
+    "cortex/blocks": 10800000,
+    "cortex/targets": 410679183,
+    "cortex/input_tokens": 17167407037,
+    "ultra_style/blocks": 10800000,
+    "ultra_style/targets": 17082628887,
+    "ultra_style/input_tokens": 17083524729,
+    "openmath/blocks": 7200000,
+    "openmath/targets": 11125217942,
+    "openmath/input_tokens": 11128924680,
+    "rewrite6/blocks": 3600000,
+    "rewrite6/targets": 5901745728,
+    "rewrite6/input_tokens": 5908216171
   },
   "sources": {
     "climbmix": {
-      "page": 3919,
-      "row": 3637,
+      "page": 3944,
+      "row": 2788,
       "start": {
-        "file": 189,
+        "file": 190,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 57,
+          "row_group": 75,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 16055861
+      "offset": 2048,
+      "documents": 16157412
     },
     "rewrite6": {
-      "page": 29,
-      "row": 4062,
+      "page": 44,
+      "row": 113,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 118784
+          "row": 180224
         }
       },
       "epoch": 44,
       "offset": 2048,
-      "documents": 9160622
+      "documents": 9218113
     },
     "openmath": {
-      "page": 2449,
-      "row": 2540,
+      "page": 2465,
+      "row": 197,
       "start": {
-        "file": 22,
+        "file": 23,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 424,
-          "row": 804
+          "row_group": 53,
+          "row": 690
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 10033644
+      "documents": 10096837
     },
     "ultra_qa": {
-      "page": 1059,
-      "row": 774,
+      "page": 1065,
+      "row": 3552,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 2,
-          "row": 51599
+          "row_group": 3,
+          "row": 11688
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 4338438
+      "documents": 4365792
     },
     "ultra_style": {
-      "page": 1616,
-      "row": 1799,
+      "page": 1626,
+      "row": 2808,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 29453
+          "row": 70413
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 6620935
+      "offset": 0,
+      "documents": 6662904
     },
     "cortex": {
-      "page": 208145,
-      "row": 16,
+      "page": 209452,
+      "row": 75,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 26642576
+      "documents": 26809931
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 795000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "c733b97c332695bc7f7abc60464b770db4d4fcc6b88b47a5682c5c028b69b18b", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "023c990bb11b42ee0e1c6492a696588ada4b8af501b7eebe7875277af2fde1e4", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d", "shard_scheduler.json": "94c6529555273c279838970e2265eb9fd5f7d996a5c9d1e791a026e9904a5f57", "coverage.json": "b9b4d36d0e722bc228f69164009d751a999eaf64d6b1a370460af2d29f35b1cc", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000795000"}
+{"step": 800000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "8dbf79dba5ff97079bc140e423cf40b0ef487b5e04e4c30f7760173437d08164", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "84eb35d156f31624644b274e41f57e3691a9338e88ee83073b01c24b4744a8cf", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d", "shard_scheduler.json": "cc275791320ae7c7d3b065796c547bbaa39e7bc94e15c07e5ba7950b8dffb861", "coverage.json": "666363f885b45a74f2b6a41ece402cf5806e0f534f2f95e854c20588602a307a", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000800000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c733b97c332695bc7f7abc60464b770db4d4fcc6b88b47a5682c5c028b69b18b
+oid sha256:8dbf79dba5ff97079bc140e423cf40b0ef487b5e04e4c30f7760173437d08164
 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": 77039.38278939603,
-      "required_pages_per_second_upper_estimate": 0.005329763257366259
+      "required_token_positions_per_second": 77170.39332101293,
+      "required_pages_per_second_upper_estimate": 0.005338826869929831
     },
     {
       "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": 9629.922848674503,
-      "required_pages_per_second_upper_estimate": 0.004031847601083585
+      "required_token_positions_per_second": 9646.299165126617,
+      "required_pages_per_second_upper_estimate": 0.004038704022805709
     },
     {
       "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": 19259.845697349007,
-      "required_pages_per_second_upper_estimate": 0.004011317942470868
+      "required_token_positions_per_second": 19292.598330253233,
+      "required_pages_per_second_upper_estimate": 0.004018139452159805
     },
     {
       "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": 28889.768546023508,
-      "required_pages_per_second_upper_estimate": 0.0017318975122117143
+      "required_token_positions_per_second": 28938.89749537985,
+      "required_pages_per_second_upper_estimate": 0.0017348427177100649
     },
     {
       "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": 28889.768546023508,
-      "required_pages_per_second_upper_estimate": 0.002246052431790947
+      "required_token_positions_per_second": 28938.89749537985,
+      "required_pages_per_second_upper_estimate": 0.0022498719915079917
     },
     {
       "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": 28889.768546023508,
-      "required_pages_per_second_upper_estimate": 0.37056693149168823
+      "required_token_positions_per_second": 28938.89749537985,
+      "required_pages_per_second_upper_estimate": 0.37119710490347546
     }
   ],
   "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": 0.4536370959920646,
-    "producer_p95_seconds": 0.2912266159983119,
-    "gpu_update_seconds": 0.9570170129938812,
-    "producer_batches_per_second": 2.204405258818368,
-    "gpu_batches_per_second": 1.0449135035454105,
-    "producer_headroom": 2.109653336222358,
+    "producer_batch_seconds": 0.1137480395858006,
+    "producer_p95_seconds": 0.34392818799824454,
+    "gpu_update_seconds": 0.9553923056126812,
+    "producer_batches_per_second": 8.79136030512153,
+    "gpu_batches_per_second": 1.0466904476048846,
+    "producer_headroom": 8.399197991381865,
     "sustainable": true,
-    "buffer_seconds": 60.29207181861452,
+    "buffer_seconds": 60.18971525359891,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 78,
+    "upload_pages_per_commit": 84,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 363779.76039098576,
-    "upload_bytes_per_second": 4912981.942445658,
+    "generation_bytes_per_second": 331531.55085523945,
+    "upload_bytes_per_second": 4593959.83944083,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 78.05866822049312,
+    "estimated_full_shards_per_hour": 71.13895315401923,
     "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": 1.2131986140011577,
-      "bytes": 10368948,
-      "pages": 40,
-      "shards": 4
-    },
-    {
-      "seconds": 0.9976378190003743,
-      "bytes": 2377649,
-      "pages": 29,
+      "seconds": 1.108235388997855,
+      "bytes": 5901441,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.268031460000202,
-      "bytes": 6225627,
-      "pages": 38,
+      "seconds": 1.0388182249989768,
+      "bytes": 6175327,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.058128027998464,
-      "bytes": 603020,
-      "pages": 18,
-      "shards": 1
-    },
-    {
-      "seconds": 1.6013217930012615,
-      "bytes": 18143583,
-      "pages": 46,
-      "shards": 5
+      "seconds": 1.7779917830011982,
+      "bytes": 7496140,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.0924186460033525,
-      "bytes": 6205338,
+      "seconds": 1.6659519919994636,
+      "bytes": 6121395,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2924744820011256,
-      "bytes": 2413023,
-      "pages": 29,
+      "seconds": 1.0959868669997377,
+      "bytes": 2315562,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0274238659985713,
-      "bytes": 6279174,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.3342183659988223,
+      "bytes": 11912121,
+      "pages": 41,
+      "shards": 3
     },
     {
-      "seconds": 1.6521351509982196,
-      "bytes": 5784763,
-      "pages": 34,
+      "seconds": 1.4696884980003233,
+      "bytes": 5989675,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.7041611679997004,
-      "bytes": 7601822,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.273706002997642,
-      "bytes": 11750945,
+      "seconds": 1.5069389330019476,
+      "bytes": 6220019,
       "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.1050900709997222,
-      "bytes": 2567501,
-      "pages": 35,
+      "seconds": 0.9491098379985488,
+      "bytes": 6427108,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2765829149975616,
-      "bytes": 9916034,
-      "pages": 36,
+      "seconds": 1.2226353290025145,
+      "bytes": 7620412,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.591446239999641,
-      "bytes": 5814050,
-      "pages": 33,
+      "seconds": 1.2109157329978188,
+      "bytes": 5386870,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.4009948579987395,
-      "bytes": 2472639,
+      "seconds": 1.3874870750005357,
+      "bytes": 11872388,
       "pages": 33,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2067919649998657,
-      "bytes": 5662165,
-      "pages": 21,
-      "shards": 2
-    },
-    {
-      "seconds": 1.381297531999735,
-      "bytes": 11693702,
-      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 1.4175854469976912,
-      "bytes": 7625416,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1141033280000556,
-      "bytes": 6156266,
+      "seconds": 1.3592714799960959,
+      "bytes": 2722837,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.90897045499878,
-      "bytes": 2264477,
-      "pages": 26,
+      "seconds": 1.3202570079956786,
+      "bytes": 6097373,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.6527253510030278,
-      "bytes": 10383315,
-      "pages": 40,
+      "seconds": 1.580753474998346,
+      "bytes": 10855257,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.1864270160003798,
-      "bytes": 6161998,
-      "pages": 33,
+      "seconds": 1.1245584639982553,
+      "bytes": 2759850,
+      "pages": 40,
       "shards": 2
     },
     {
-      "seconds": 1.1574043389991857,
-      "bytes": 8408946,
+      "seconds": 1.2774146450028638,
+      "bytes": 6152659,
       "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.128104429000814,
-      "bytes": 5797209,
-      "pages": 26,
       "shards": 2
     },
     {
-      "seconds": 1.344746707000013,
-      "bytes": 12817447,
-      "pages": 46,
+      "seconds": 1.6375758160065743,
+      "bytes": 13187614,
+      "pages": 36,
       "shards": 4
     },
     {
-      "seconds": 1.0572433240013197,
-      "bytes": 864410,
-      "pages": 27,
+      "seconds": 1.1867789399984758,
+      "bytes": 1047762,
+      "pages": 32,
       "shards": 1
     },
     {
-      "seconds": 1.8016076449966931,
-      "bytes": 5772859,
-      "pages": 28,
+      "seconds": 2.737723589001689,
+      "bytes": 9866131,
+      "pages": 36,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0393312469968805,
+      "bytes": 2566382,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 3.847735643001215,
-      "bytes": 2775913,
+      "seconds": 1.564457227003004,
+      "bytes": 6307452,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.592129067001224,
-      "bytes": 15451622,
-      "pages": 34,
+      "seconds": 1.4970643039996503,
+      "bytes": 11855130,
+      "pages": 30,
       "shards": 4
     },
     {
-      "seconds": 1.0286604469984013,
-      "bytes": 6223313,
-      "pages": 38,
+      "seconds": 1.1368510719985352,
+      "bytes": 7197379,
+      "pages": 42,
       "shards": 2
     },
     {
-      "seconds": 1.2432226130003983,
-      "bytes": 2637387,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.1000944999977946,
+      "bytes": 5635539,
+      "pages": 19,
+      "shards": 2
     },
     {
-      "seconds": 1.4529079339990858,
-      "bytes": 5933529,
+      "seconds": 1.2645733180033858,
+      "bytes": 2430189,
       "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2705231469990395,
-      "bytes": 7122874,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2547058620002645,
-      "bytes": 6059042,
-      "pages": 36,
+      "seconds": 1.1291347480000695,
+      "bytes": 6037077,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.5521136860006663,
-      "bytes": 11684886,
-      "pages": 33,
+      "seconds": 1.49954066099599,
+      "bytes": 11232631,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.2612271040015912,
-      "bytes": 2529037,
-      "pages": 34,
+      "seconds": 1.3497194699957618,
+      "bytes": 2429729,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.3205864709998423,
-      "bytes": 10433001,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.1435100689996034,
+      "bytes": 625562,
+      "pages": 19,
+      "shards": 1
     },
     {
-      "seconds": 1.281639933000406,
-      "bytes": 7384470,
-      "pages": 31,
+      "seconds": 1.4279748699991615,
+      "bytes": 13559996,
+      "pages": 51,
+      "shards": 4
+    },
+    {
+      "seconds": 1.198663317001774,
+      "bytes": 9767212,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.4279891010010033,
-      "bytes": 1106558,
-      "pages": 34,
+      "seconds": 1.4251030520026688,
+      "bytes": 1167175,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 2.285999409003125,
-      "bytes": 10525684,
-      "pages": 32,
+      "seconds": 1.788670624999213,
+      "bytes": 7398476,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.1848097770016466,
-      "bytes": 8443913,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.539240851001523,
+      "bytes": 6176798,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.108235388997855,
-      "bytes": 5901441,
-      "pages": 28,
+      "seconds": 1.0939249449947965,
+      "bytes": 5048949,
+      "pages": 21,
       "shards": 2
     },
     {
-      "seconds": 1.0388182249989768,
-      "bytes": 6175327,
+      "seconds": 1.358410709995951,
+      "bytes": 13879443,
+      "pages": 50,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1252228140001534,
+      "bytes": 887251,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.1445283560024109,
+      "bytes": 7481150,
       "pages": 37,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.7779917830011982,
-      "bytes": 7496140,
-      "pages": 33,
+      "seconds": 2.095004743998288,
+      "bytes": 10399963,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.6659519919994636,
-      "bytes": 6121395,
-      "pages": 38,
+      "seconds": 0.994942393997917,
+      "bytes": 1292130,
+      "pages": 36,
+      "shards": 1
+    },
+    {
+      "seconds": 1.574394162002136,
+      "bytes": 12830257,
+      "pages": 24,
+      "shards": 4
+    },
+    {
+      "seconds": 1.6963619910020498,
+      "bytes": 6365269,
+      "pages": 45,
       "shards": 2
     },
     {
-      "seconds": 1.0959868669997377,
-      "bytes": 2315562,
-      "pages": 28,
+      "seconds": 1.0425699910047115,
+      "bytes": 4497395,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.3342183659988223,
-      "bytes": 11912121,
-      "pages": 41,
+      "seconds": 1.688526213998557,
+      "bytes": 7586011,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.4696884980003233,
-      "bytes": 5989675,
-      "pages": 29,
+      "seconds": 1.5618516569957137,
+      "bytes": 5919633,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.5069389330019476,
-      "bytes": 6220019,
-      "pages": 34,
+      "seconds": 2.862179110001307,
+      "bytes": 6924568,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9491098379985488,
-      "bytes": 6427108,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.8365363409975544,
+      "bytes": 11520166,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.2226353290025145,
-      "bytes": 7620412,
+      "seconds": 1.1648409480039845,
+      "bytes": 7473766,
       "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.2109157329978188,
-      "bytes": 5386870,
-      "pages": 32,
-      "shards": 2
+      "seconds": 1.0564176369953202,
+      "bytes": 870650,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.3874870750005357,
-      "bytes": 11872388,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.4489900529952138,
+      "bytes": 10666654,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.3592714799960959,
-      "bytes": 2722837,
-      "pages": 36,
+      "seconds": 1.0202623110017157,
+      "bytes": 2376232,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.3202570079956786,
-      "bytes": 6097373,
-      "pages": 30,
+      "seconds": 1.3351656059967354,
+      "bytes": 6483786,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.580753474998346,
-      "bytes": 10855257,
-      "pages": 30,
+      "seconds": 1.8462947400039411,
+      "bytes": 11598158,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.1245584639982553,
-      "bytes": 2759850,
-      "pages": 40,
+      "seconds": 3.023687531000178,
+      "bytes": 6495633,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.523345616995357,
+      "bytes": 6337115,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2774146450028638,
-      "bytes": 6152659,
-      "pages": 37,
+      "seconds": 1.3345436189993052,
+      "bytes": 2354584,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.6375758160065743,
-      "bytes": 13187614,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.836964503992931,
+      "bytes": 6252743,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1867789399984758,
-      "bytes": 1047762,
+      "seconds": 2.1564883190003457,
+      "bytes": 10319080,
       "pages": 32,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 2.737723589001689,
-      "bytes": 9866131,
+      "seconds": 1.3421651660028147,
+      "bytes": 8322765,
       "pages": 36,
-      "shards": 4
+      "shards": 3
     },
     {
-      "seconds": 1.0393312469968805,
-      "bytes": 2566382,
-      "pages": 31,
+      "seconds": 1.1529908149968833,
+      "bytes": 5931904,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.564457227003004,
-      "bytes": 6307452,
-      "pages": 38,
+      "seconds": 1.4366694560012547,
+      "bytes": 7813323,
+      "pages": 41,
+      "shards": 3
+    },
+    {
+      "seconds": 1.6668840290003573,
+      "bytes": 4874073,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.4970643039996503,
-      "bytes": 11855130,
-      "pages": 30,
-      "shards": 4
+      "seconds": 1.9653244989967789,
+      "bytes": 6133990,
+      "pages": 38,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:023c990bb11b42ee0e1c6492a696588ada4b8af501b7eebe7875277af2fde1e4
-size 721788287
+oid sha256:84eb35d156f31624644b274e41f57e3691a9338e88ee83073b01c24b4744a8cf
+size 721788351