Qwen90M step 240000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 237163,
+  "step": 240000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "2b20b64846b5ec5f1a1266ad70e66bdcf4161c507eb849f41d4457b347e1fef8",
+    "model.safetensors": "65898f02c82a8377d9bda1d5b03ff50aeed1638370f2d8e06f763970f456a634",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "3817012a1ce6a6ac292529de34664fef83f2285d26044dba167d3aaa20957a39",
+    "trainer.pt": "93447def639b9eb557ef085795282a30d7afe7ce82e2b6d060081b6e8ae7533d",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
-    "performance.json": "7f72f0a0e868a4baec06f238f922d0c5d7e397a51386c11ac32acc776d4d3522",
-    "shard_scheduler.json": "7642e374dd789aa75440578f87a3db0e21df171d6322760efedf97e6eb520c03",
-    "coverage.json": "70e3c6c517ccfd837f7cb477fd294e1034ff341fbf2983c3135f90eb9c011c05",
+    "performance.json": "ac15a69eb79c2ac9ef8406edc12d9ae8abafb561b0f74b54b23d5920ee4fe3f7",
+    "shard_scheduler.json": "2c0ed3cddd320bf86cf19e2dddc35b552898c1782e20beeea5fb498238b66008",
+    "coverage.json": "aafdacf1fbbe0595d411cba8e2eec7aa9696122beadc4ad6618a4d163528507a",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 237163,
+  "step": 240000,
   "stats": {
-    "climbmix/blocks": 8537867,
-    "climbmix/targets": 14271950160,
-    "climbmix/input_tokens": 14272854649,
-    "ultra_qa/blocks": 3201701,
-    "ultra_qa/targets": 5223297877,
-    "ultra_qa/input_tokens": 5223307978,
-    "cortex/blocks": 3201700,
-    "cortex/targets": 121802303,
-    "cortex/input_tokens": 5089120155,
-    "ultra_style/blocks": 3201701,
-    "ultra_style/targets": 5064464924,
-    "ultra_style/input_tokens": 5064730515,
-    "openmath/blocks": 2134467,
-    "openmath/targets": 3298047515,
-    "openmath/input_tokens": 3299146987,
-    "rewrite6/blocks": 1067234,
-    "rewrite6/targets": 1749558797,
-    "rewrite6/input_tokens": 1751477016
+    "climbmix/blocks": 8640000,
+    "climbmix/targets": 14442418448,
+    "climbmix/input_tokens": 14443333774,
+    "ultra_qa/blocks": 3240000,
+    "ultra_qa/targets": 5285705457,
+    "ultra_qa/input_tokens": 5285715652,
+    "cortex/blocks": 3240000,
+    "cortex/targets": 123256375,
+    "cortex/input_tokens": 5149964923,
+    "ultra_style/blocks": 3240000,
+    "ultra_style/targets": 5125067906,
+    "ultra_style/input_tokens": 5125336525,
+    "openmath/blocks": 2160000,
+    "openmath/targets": 3337488563,
+    "openmath/input_tokens": 3338601175,
+    "rewrite6/blocks": 1080000,
+    "rewrite6/targets": 1770500519,
+    "rewrite6/input_tokens": 1772442074
   },
   "sources": {
     "climbmix": {
-      "page": 1169,
-      "row": 1585,
+      "page": 1183,
+      "row": 1723,
       "start": {
-        "file": 56,
+        "file": 57,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 48,
+          "row_group": 21,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 4789809
+      "documents": 4847291
     },
     "rewrite6": {
-      "page": 15,
-      "row": 1043,
+      "page": 23,
+      "row": 1439,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 61440
+          "row": 94208
         }
       },
       "epoch": 13,
       "offset": 0,
-      "documents": 2732735
+      "documents": 2765899
     },
     "openmath": {
-      "page": 730,
-      "row": 4073,
+      "page": 739,
+      "row": 3019,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 370,
-          "row": 180
+          "row_group": 407,
+          "row": 44
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 2994153
+      "documents": 3029963
     },
     "ultra_qa": {
-      "page": 315,
-      "row": 4076,
+      "page": 319,
+      "row": 3147,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 57255
+          "row_group": 4,
+          "row": 9056
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 1294316
+      "offset": 4096,
+      "documents": 1309771
     },
     "ultra_style": {
-      "page": 482,
-      "row": 1439,
+      "page": 488,
+      "row": 260,
       "start": {
         "file": 2,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 8799
+          "row": 33375
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 1975711
+      "documents": 1999108
     },
     "cortex": {
-      "page": 62090,
-      "row": 107,
+      "page": 62833,
+      "row": 5,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 7947627
+      "documents": 8042629
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 237163, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "2b20b64846b5ec5f1a1266ad70e66bdcf4161c507eb849f41d4457b347e1fef8", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "3817012a1ce6a6ac292529de34664fef83f2285d26044dba167d3aaa20957a39", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "7f72f0a0e868a4baec06f238f922d0c5d7e397a51386c11ac32acc776d4d3522", "shard_scheduler.json": "7642e374dd789aa75440578f87a3db0e21df171d6322760efedf97e6eb520c03", "coverage.json": "70e3c6c517ccfd837f7cb477fd294e1034ff341fbf2983c3135f90eb9c011c05", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000237163"}
+{"step": 240000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "65898f02c82a8377d9bda1d5b03ff50aeed1638370f2d8e06f763970f456a634", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "93447def639b9eb557ef085795282a30d7afe7ce82e2b6d060081b6e8ae7533d", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "ac15a69eb79c2ac9ef8406edc12d9ae8abafb561b0f74b54b23d5920ee4fe3f7", "shard_scheduler.json": "2c0ed3cddd320bf86cf19e2dddc35b552898c1782e20beeea5fb498238b66008", "coverage.json": "aafdacf1fbbe0595d411cba8e2eec7aa9696122beadc4ad6618a4d163528507a", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000240000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2b20b64846b5ec5f1a1266ad70e66bdcf4161c507eb849f41d4457b347e1fef8
+oid sha256:65898f02c82a8377d9bda1d5b03ff50aeed1638370f2d8e06f763970f456a634
 size 360622744

performance.json

@@ -49,11 +49,11 @@
   "selected": {
     "microbatch": 86,
     "mode": "compiled",
-    "microbatch_seconds": 0.9016621779999809,
+    "microbatch_seconds": 0.8988038599999868,
     "peak_gb": 85.39913082122803,
     "fits_budget": true,
-    "estimated_update_seconds": 0.9578150019999612,
-    "input_tokens_per_second": 195337.01678679453
+    "estimated_update_seconds": 0.9552281219999941,
+    "input_tokens_per_second": 195958.21495470946
   },
   "measurements": [
     {

shard_scheduler.json

@@ -22,8 +22,8 @@
       "tokens_per_second": 1062967.6168812835,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 75472.3444404108,
-      "required_pages_per_second_upper_estimate": 0.006541751533186617
+      "required_token_positions_per_second": 75700.32285632046,
+      "required_pages_per_second_upper_estimate": 0.006561512124471654
     },
     {
       "source": "rewrite6",
@@ -36,8 +36,8 @@
       "tokens_per_second": 1206672.263377472,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 9434.04305505135,
-      "required_pages_per_second_upper_estimate": 0.0037633808261733487
+      "required_token_positions_per_second": 9462.540357040058,
+      "required_pages_per_second_upper_estimate": 0.003774748826009278
     },
     {
       "source": "openmath",
@@ -50,8 +50,8 @@
       "tokens_per_second": 410042.7581134458,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 18868.0861101027,
-      "required_pages_per_second_upper_estimate": 0.004364799654598218
+      "required_token_positions_per_second": 18925.080714080115,
+      "required_pages_per_second_upper_estimate": 0.004377984353157622
     },
     {
       "source": "ultra_qa",
@@ -64,8 +64,8 @@
       "tokens_per_second": 2163486.5624068812,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28302.129165154052,
-      "required_pages_per_second_upper_estimate": 0.0017234511763620484
+      "required_token_positions_per_second": 28387.62107112017,
+      "required_pages_per_second_upper_estimate": 0.0017286571848940193
     },
     {
       "source": "ultra_style",
@@ -78,8 +78,8 @@
       "tokens_per_second": 1546694.2737633225,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28302.129165154052,
-      "required_pages_per_second_upper_estimate": 0.002178776248441411
+      "required_token_positions_per_second": 28387.62107112017,
+      "required_pages_per_second_upper_estimate": 0.002185357652019425
     },
     {
       "source": "cortex",
@@ -92,8 +92,8 @@
       "tokens_per_second": 1015177.8395898147,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 22,
-      "required_token_positions_per_second": 28302.129165154052,
-      "required_pages_per_second_upper_estimate": 0.35141772309812946
+      "required_token_positions_per_second": 28387.62107112017,
+      "required_pages_per_second_upper_estimate": 0.3524792464472251
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -157,410 +157,170 @@
   "selected": {
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
-    "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.12409712750809376,
-    "producer_p95_seconds": 0.4301016780009377,
-    "gpu_update_seconds": 0.9768876340950552,
-    "producer_batches_per_second": 8.05820424759452,
-    "gpu_batches_per_second": 1.0236591856609538,
-    "producer_headroom": 7.871960082487336,
+    "batch_memory_estimate_bytes": 690670,
+    "producer_batch_seconds": 0.2644437009218734,
+    "producer_p95_seconds": 0.16535117300009006,
+    "gpu_update_seconds": 0.9739456480249902,
+    "producer_batches_per_second": 3.781523237323915,
+    "gpu_batches_per_second": 1.026751340824659,
+    "producer_headroom": 3.6829980998969996,
     "sustainable": true,
-    "buffer_seconds": 60.56703331389342,
+    "buffer_seconds": 60.38463017754939,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 84,
+    "upload_pages_per_commit": 83,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 379406.6242124819,
-    "upload_bytes_per_second": 5575493.562052467,
+    "generation_bytes_per_second": 516494.2735893789,
+    "upload_bytes_per_second": 4500469.795094228,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 81.41182942181437,
+    "estimated_full_shards_per_hour": 110.82764774094606,
     "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": 1.3212683820020175,
-      "bytes": 7578218,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9010331350000342,
-      "bytes": 973266,
-      "pages": 27,
+      "seconds": 0.8557939730000044,
+      "bytes": 46834,
+      "pages": 1,
       "shards": 1
     },
     {
-      "seconds": 1.851456552001764,
-      "bytes": 16129409,
-      "pages": 39,
-      "shards": 4
-    },
-    {
-      "seconds": 1.138110815998516,
-      "bytes": 2483869,
-      "pages": 32,
-      "shards": 2
-    },
-    {
-      "seconds": 1.5748863410044578,
-      "bytes": 6345410,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1974535199988168,
-      "bytes": 6123455,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.302649115998065,
-      "bytes": 6509625,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.073786435998045,
-      "bytes": 5871200,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9513727629964706,
-      "bytes": 2591189,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2582520649957587,
-      "bytes": 11695145,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 1.135505132995604,
-      "bytes": 10769907,
-      "pages": 36,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0033940869980142,
-      "bytes": 2504989,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0786675720009953,
-      "bytes": 5961901,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.8719808540045051,
-      "bytes": 2734178,
-      "pages": 38,
+      "seconds": 1.4821450189999723,
+      "bytes": 5748205,
+      "pages": 24,
       "shards": 2
     },
     {
-      "seconds": 1.0888003759973799,
-      "bytes": 5559059,
-      "pages": 23,
-      "shards": 2
+      "seconds": 1.3858465779999278,
+      "bytes": 775414,
+      "pages": 24,
+      "shards": 1
     },
     {
-      "seconds": 1.564358691997768,
-      "bytes": 15793806,
-      "pages": 43,
+      "seconds": 1.4328443340000376,
+      "bytes": 11483713,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 0.8641434079981991,
-      "bytes": 2665706,
-      "pages": 37,
-      "shards": 2
+      "seconds": 0.909485917999973,
+      "bytes": 1076059,
+      "pages": 32,
+      "shards": 1
     },
     {
-      "seconds": 0.9742672610009322,
-      "bytes": 6080743,
-      "pages": 28,
+      "seconds": 0.961313237000013,
+      "bytes": 6243214,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1234689760021865,
-      "bytes": 10516427,
-      "pages": 39,
+      "seconds": 1.4556228910000755,
+      "bytes": 7655201,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.2163533749990165,
-      "bytes": 2238236,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9348791060037911,
-      "bytes": 6426893,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.129189448998659,
-      "bytes": 8174843,
-      "pages": 30,
+      "seconds": 1.1608069779999823,
+      "bytes": 11166576,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.1518778129975544,
-      "bytes": 10667449,
-      "pages": 40,
+      "seconds": 1.0247551580000618,
+      "bytes": 7489968,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.092219940997893,
-      "bytes": 846588,
-      "pages": 27,
+      "seconds": 0.7696208980000847,
+      "bytes": 1062863,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.1457775259987102,
-      "bytes": 7463584,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 0.950148095995246,
-      "bytes": 5837048,
-      "pages": 28,
+      "seconds": 1.0704197729999123,
+      "bytes": 6206676,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 0.8334624849958345,
-      "bytes": 871180,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.5179614139997284,
-      "bytes": 17372933,
-      "pages": 43,
-      "shards": 5
-    },
-    {
-      "seconds": 0.9750086150015704,
-      "bytes": 928101,
-      "pages": 28,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0726397860053112,
-      "bytes": 7787197,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9909355400013737,
-      "bytes": 5704571,
-      "pages": 21,
+      "seconds": 2.8963105739999264,
+      "bytes": 7237253,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.0439709420024883,
-      "bytes": 6002810,
-      "pages": 44,
+      "seconds": 1.3083305300001484,
+      "bytes": 5580567,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.7808462090033572,
-      "bytes": 7399793,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2099171500012744,
-      "bytes": 12004299,
-      "pages": 41,
+      "seconds": 1.1845111140000881,
+      "bytes": 11699180,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.7972624179965351,
-      "bytes": 919254,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.5076305830007186,
-      "bytes": 11558434,
-      "pages": 35,
-      "shards": 4
-    },
-    {
-      "seconds": 0.8286585169989849,
-      "bytes": 1181916,
+      "seconds": 0.859392975999981,
+      "bytes": 2445063,
       "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 1.184618612001941,
-      "bytes": 7236384,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0676331659997231,
-      "bytes": 5948596,
-      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.030731860002561,
-      "bytes": 4760298,
-      "pages": 25,
-      "shards": 2
-    },
-    {
-      "seconds": 1.582079469000746,
-      "bytes": 13247311,
-      "pages": 40,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0703849589990568,
-      "bytes": 1241496,
-      "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 1.07058091199724,
-      "bytes": 5784118,
-      "pages": 26,
+      "seconds": 1.0558128229999966,
+      "bytes": 6261983,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 2.4059382019986515,
-      "bytes": 12510715,
-      "pages": 47,
+      "seconds": 1.3157395409998571,
+      "bytes": 12201352,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 0.7931236949953018,
-      "bytes": 1028533,
-      "pages": 27,
+      "seconds": 2.8152504160002536,
+      "bytes": 1180754,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 2.0690407029978815,
-      "bytes": 7202195,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 1.457639926004049,
-      "bytes": 7115441,
-      "pages": 39,
-      "shards": 2
-    },
-    {
-      "seconds": 1.29948361699644,
-      "bytes": 9783282,
-      "pages": 32,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0601626389980083,
-      "bytes": 7547567,
-      "pages": 29,
-      "shards": 3
-    },
-    {
-      "seconds": 1.118395095996675,
-      "bytes": 1248062,
+      "seconds": 1.0497820410000713,
+      "bytes": 5934890,
       "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 0.9650491309948848,
-      "bytes": 5725695,
-      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1484884869933012,
-      "bytes": 6640918,
-      "pages": 32,
+      "seconds": 1.250543007000033,
+      "bytes": 8195692,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.1328601960049127,
-      "bytes": 11867525,
+      "seconds": 1.6225744519997534,
+      "bytes": 9437094,
       "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0216524100032984,
-      "bytes": 7553152,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 0.8235062559979269,
-      "bytes": 854296,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.2072945199979586,
-      "bytes": 10797475,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.3132626469960087,
-      "bytes": 2377167,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.8975952609980595,
-      "bytes": 5947609,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3013317000004463,
-      "bytes": 11518633,
+      "seconds": 1.1101808009998422,
+      "bytes": 6059135,
       "pages": 35,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1227065580023918,
-      "bytes": 6254929,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.5046920809982112,
-      "bytes": 5774998,
-      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 0.8562071050037048,
-      "bytes": 2686484,
-      "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2383149749948643,
-      "bytes": 6204112,
-      "pages": 30,
+      "seconds": 1.0066997960002482,
+      "bytes": 2574525,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.4738188229966909,
-      "bytes": 10769021,
-      "pages": 39,
+      "seconds": 3.0401188460000412,
+      "bytes": 10860861,
+      "pages": 35,
       "shards": 3
     }
   ],
@@ -606,7 +366,7 @@
   },
   "cache_reused": true,
   "resume_probe_seconds": [
-    3.044028898000022,
-    0.01595121099995822
+    2.844574209999962,
+    0.2503580399999805
   ]
 }

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3817012a1ce6a6ac292529de34664fef83f2285d26044dba167d3aaa20957a39
-size 721779903
+oid sha256:93447def639b9eb557ef085795282a30d7afe7ce82e2b6d060081b6e8ae7533d
+size 721779519