Qwen90M step 805000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 800000,
+  "step": 805000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "8dbf79dba5ff97079bc140e423cf40b0ef487b5e04e4c30f7760173437d08164",
+    "model.safetensors": "f81353344f2caa3305d0563c6e2992bc9eca917fdb0f44f55a65bbf690de87f9",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "84eb35d156f31624644b274e41f57e3691a9338e88ee83073b01c24b4744a8cf",
+    "trainer.pt": "1232789bccab1559e521ddbb828d7a33d03698d8a2d36aeb0341e6a2e27d4c92",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
-    "performance.json": "d91455eeac02a2960a60e3a2edc7d99796576b62edb422642a8beb3b824ff02d",
-    "shard_scheduler.json": "cc275791320ae7c7d3b065796c547bbaa39e7bc94e15c07e5ba7950b8dffb861",
-    "coverage.json": "666363f885b45a74f2b6a41ece402cf5806e0f534f2f95e854c20588602a307a",
+    "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7",
+    "shard_scheduler.json": "671cb6a206e18282c9946eb8a1b247f024ce1af7927d526a978429fd995d40c4",
+    "coverage.json": "413a5d11bd5c4ca98af7d6cdbc7f9d3951a2feb0417fce131129dc6f92afcc10",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 800000,
+  "step": 805000,
   "stats": {
-    "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
+    "climbmix/blocks": 28980000,
+    "climbmix/targets": 48444394688,
+    "climbmix/input_tokens": 48447464623,
+    "ultra_qa/blocks": 10867500,
+    "ultra_qa/targets": 17729141745,
+    "ultra_qa/input_tokens": 17729175912,
+    "cortex/blocks": 10867500,
+    "cortex/targets": 413246363,
+    "cortex/input_tokens": 17274632186,
+    "ultra_style/blocks": 10867500,
+    "ultra_style/targets": 17189552467,
+    "ultra_style/input_tokens": 17190453888,
+    "openmath/blocks": 7245000,
+    "openmath/targets": 11194750151,
+    "openmath/input_tokens": 11198479888,
+    "rewrite6/blocks": 3622500,
+    "rewrite6/targets": 5938628484,
+    "rewrite6/input_tokens": 5945139165
   },
   "sources": {
     "climbmix": {
-      "page": 3944,
-      "row": 2788,
+      "page": 3969,
+      "row": 558,
       "start": {
-        "file": 190,
+        "file": 192,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 75,
+          "row_group": 9,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 16157412
+      "documents": 16257582
     },
     "rewrite6": {
-      "page": 44,
-      "row": 113,
+      "page": 7,
+      "row": 3564,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 180224
+          "row": 28672
         }
       },
-      "epoch": 44,
-      "offset": 2048,
-      "documents": 9218113
+      "epoch": 45,
+      "offset": 0,
+      "documents": 9275416
     },
     "openmath": {
-      "page": 2465,
-      "row": 197,
+      "page": 2480,
+      "row": 1695,
       "start": {
         "file": 23,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 53,
-          "row": 690
+          "row_group": 115,
+          "row": 130
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 10096837
+      "documents": 10159775
     },
     "ultra_qa": {
-      "page": 1065,
-      "row": 3552,
+      "page": 1072,
+      "row": 2161,
       "start": {
         "file": 8,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 3,
-          "row": 11688
+          "row": 40360
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 4365792
+      "documents": 4393073
     },
     "ultra_style": {
-      "page": 1626,
-      "row": 2808,
+      "page": 1636,
+      "row": 3207,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 5,
-          "row": 70413
+          "row_group": 6,
+          "row": 27157
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 6662904
+      "offset": 2048,
+      "documents": 6704263
     },
     "cortex": {
-      "page": 209452,
-      "row": 75,
+      "page": 210760,
+      "row": 21,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 26809931
+      "documents": 26977301
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"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"}
+{"step": 805000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "f81353344f2caa3305d0563c6e2992bc9eca917fdb0f44f55a65bbf690de87f9", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "1232789bccab1559e521ddbb828d7a33d03698d8a2d36aeb0341e6a2e27d4c92", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "eb45b2d9ecb051b6b7e592c18f2a0bfebf4278147b783ba131f7a9a4453a9ec7", "shard_scheduler.json": "671cb6a206e18282c9946eb8a1b247f024ce1af7927d526a978429fd995d40c4", "coverage.json": "413a5d11bd5c4ca98af7d6cdbc7f9d3951a2feb0417fce131129dc6f92afcc10", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000805000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8dbf79dba5ff97079bc140e423cf40b0ef487b5e04e4c30f7760173437d08164
+oid sha256:f81353344f2caa3305d0563c6e2992bc9eca917fdb0f44f55a65bbf690de87f9
 size 360622744

performance.json

@@ -49,11 +49,11 @@
   "selected": {
     "microbatch": 90,
     "mode": "compiled",
-    "microbatch_seconds": 0.9481044159999783,
+    "microbatch_seconds": 0.9411944559999483,
     "peak_gb": 89.37790203094482,
     "fits_budget": true,
-    "estimated_update_seconds": 0.9481044159999783,
-    "input_tokens_per_second": 194408.96687059014
+    "estimated_update_seconds": 0.9411944559999483,
+    "input_tokens_per_second": 195836.25766704488
   },
   "measurements": [
     {

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": 77170.39332101293,
-      "required_pages_per_second_upper_estimate": 0.005338826869929831
+      "required_token_positions_per_second": 76909.91644078335,
+      "required_pages_per_second_upper_estimate": 0.005320806474965917
     },
     {
       "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": 9646.299165126617,
-      "required_pages_per_second_upper_estimate": 0.004038704022805709
+      "required_token_positions_per_second": 9613.739555097918,
+      "required_pages_per_second_upper_estimate": 0.004025071994008668
     },
     {
       "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": 19292.598330253233,
-      "required_pages_per_second_upper_estimate": 0.004018139452159805
+      "required_token_positions_per_second": 19227.479110195836,
+      "required_pages_per_second_upper_estimate": 0.004004576835931166
     },
     {
       "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": 28938.89749537985,
-      "required_pages_per_second_upper_estimate": 0.0017348427177100649
+      "required_token_positions_per_second": 28841.218665293753,
+      "required_pages_per_second_upper_estimate": 0.001728987020993341
     },
     {
       "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": 28938.89749537985,
-      "required_pages_per_second_upper_estimate": 0.0022498719915079917
+      "required_token_positions_per_second": 28841.218665293753,
+      "required_pages_per_second_upper_estimate": 0.0022422778921125652
     },
     {
       "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": 28938.89749537985,
-      "required_pages_per_second_upper_estimate": 0.37119710490347546
+      "required_token_positions_per_second": 28841.218665293753,
+      "required_pages_per_second_upper_estimate": 0.36994418575048743
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,73 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "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,
+    "producer_batch_seconds": 0.0739909046796825,
+    "producer_p95_seconds": 0.15531819999978325,
+    "gpu_update_seconds": 0.9586280080900458,
+    "producer_batches_per_second": 13.515174660036216,
+    "gpu_batches_per_second": 1.0431575038083678,
+    "producer_headroom": 12.95602496333958,
     "sustainable": true,
-    "buffer_seconds": 60.18971525359891,
+    "buffer_seconds": 60.393564509672885,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 84,
+    "upload_pages_per_commit": 78,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 331531.55085523945,
-    "upload_bytes_per_second": 4593959.83944083,
+    "generation_bytes_per_second": 383905.7653317095,
+    "upload_bytes_per_second": 3532908.8334232424,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 71.13895315401923,
+    "estimated_full_shards_per_hour": 82.37724037135567,
     "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": 1.108235388997855,
-      "bytes": 5901441,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0388182249989768,
-      "bytes": 6175327,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.7779917830011982,
-      "bytes": 7496140,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 1.6659519919994636,
-      "bytes": 6121395,
-      "pages": 38,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0959868669997377,
-      "bytes": 2315562,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3342183659988223,
-      "bytes": 11912121,
-      "pages": 41,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4696884980003233,
-      "bytes": 5989675,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.5069389330019476,
-      "bytes": 6220019,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9491098379985488,
-      "bytes": 6427108,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.6723634379995929,
+      "bytes": 50251,
+      "pages": 1,
+      "shards": 1
     },
     {
-      "seconds": 1.2226353290025145,
-      "bytes": 7620412,
+      "seconds": 2.1410375759996896,
+      "bytes": 10006036,
       "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.2109157329978188,
-      "bytes": 5386870,
-      "pages": 32,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3874870750005357,
-      "bytes": 11872388,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 1.3592714799960959,
-      "bytes": 2722837,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3202570079956786,
-      "bytes": 6097373,
+      "seconds": 1.5198384699997405,
+      "bytes": 2430008,
       "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.580753474998346,
-      "bytes": 10855257,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1245584639982553,
-      "bytes": 2759850,
-      "pages": 40,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2774146450028638,
-      "bytes": 6152659,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.6375758160065743,
-      "bytes": 13187614,
-      "pages": 36,
-      "shards": 4
-    },
-    {
-      "seconds": 1.1867789399984758,
-      "bytes": 1047762,
-      "pages": 32,
-      "shards": 1
-    },
-    {
-      "seconds": 2.737723589001689,
-      "bytes": 9866131,
+      "seconds": 1.2644712469991646,
+      "bytes": 6316580,
       "pages": 36,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0393312469968805,
-      "bytes": 2566382,
-      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.564457227003004,
-      "bytes": 6307452,
+      "seconds": 1.6103071739998995,
+      "bytes": 7727369,
       "pages": 38,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4970643039996503,
-      "bytes": 11855130,
-      "pages": 30,
-      "shards": 4
-    },
-    {
-      "seconds": 1.1368510719985352,
-      "bytes": 7197379,
-      "pages": 42,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1000944999977946,
-      "bytes": 5635539,
-      "pages": 19,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2645733180033858,
-      "bytes": 2430189,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1291347480000695,
-      "bytes": 6037077,
-      "pages": 34,
-      "shards": 2
-    },
-    {
-      "seconds": 1.49954066099599,
-      "bytes": 11232631,
-      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.3497194699957618,
-      "bytes": 2429729,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1435100689996034,
-      "bytes": 625562,
-      "pages": 19,
-      "shards": 1
-    },
-    {
-      "seconds": 1.4279748699991615,
-      "bytes": 13559996,
-      "pages": 51,
-      "shards": 4
-    },
-    {
-      "seconds": 1.198663317001774,
-      "bytes": 9767212,
-      "pages": 29,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4251030520026688,
-      "bytes": 1167175,
-      "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 1.788670624999213,
-      "bytes": 7398476,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.539240851001523,
-      "bytes": 6176798,
-      "pages": 38,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0939249449947965,
-      "bytes": 5048949,
-      "pages": 21,
-      "shards": 2
-    },
-    {
-      "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": 3
-    },
-    {
-      "seconds": 2.095004743998288,
-      "bytes": 10399963,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "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": 1.0425699910047115,
-      "bytes": 4497395,
-      "pages": 30,
-      "shards": 2
-    },
-    {
-      "seconds": 1.688526213998557,
-      "bytes": 7586011,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.5618516569957137,
-      "bytes": 5919633,
-      "pages": 30,
-      "shards": 2
-    },
-    {
-      "seconds": 2.862179110001307,
-      "bytes": 6924568,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.8365363409975544,
-      "bytes": 11520166,
-      "pages": 32,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1648409480039845,
-      "bytes": 7473766,
-      "pages": 40,
+      "seconds": 1.5575974499997756,
+      "bytes": 11152630,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.0564176369953202,
-      "bytes": 870650,
+      "seconds": 1.9014979050007241,
+      "bytes": 5893171,
       "pages": 27,
-      "shards": 1
-    },
-    {
-      "seconds": 1.4489900529952138,
-      "bytes": 10666654,
-      "pages": 41,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0202623110017157,
-      "bytes": 2376232,
-      "pages": 30,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3351656059967354,
-      "bytes": 6483786,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.8462947400039411,
-      "bytes": 11598158,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 3.023687531000178,
-      "bytes": 6495633,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.523345616995357,
-      "bytes": 6337115,
-      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.3345436189993052,
-      "bytes": 2354584,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.836964503992931,
-      "bytes": 6252743,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 2.1564883190003457,
-      "bytes": 10319080,
-      "pages": 32,
-      "shards": 3
-    },
-    {
-      "seconds": 1.3421651660028147,
-      "bytes": 8322765,
+      "seconds": 1.375044964999688,
+      "bytes": 2500711,
       "pages": 36,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1529908149968833,
-      "bytes": 5931904,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.4366694560012547,
-      "bytes": 7813323,
-      "pages": 41,
-      "shards": 3
-    },
-    {
-      "seconds": 1.6668840290003573,
-      "bytes": 4874073,
-      "pages": 32,
-      "shards": 2
-    },
-    {
-      "seconds": 1.9653244989967789,
-      "bytes": 6133990,
-      "pages": 38,
       "shards": 2
     }
   ],
@@ -606,7 +270,7 @@
   },
   "cache_reused": true,
   "resume_probe_seconds": [
-    3.0803219919999947,
-    0.10207374499998423
+    3.328129581999974,
+    0.10924991900014902
   ]
 }

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:84eb35d156f31624644b274e41f57e3691a9338e88ee83073b01c24b4744a8cf
-size 721788351
+oid sha256:1232789bccab1559e521ddbb828d7a33d03698d8a2d36aeb0341e6a2e27d4c92
+size 721769471