Qwen90M step 560000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 557164,
+  "step": 560000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "9aeb84d48b1a3e7876ab13cf2a9db637ae3eea5a8e14c759a56dfd289a32ce86",
+    "model.safetensors": "9fc6ec90f620abfbf95bf8c6ae75a133a76b595b55c08435360d181ac4591f3c",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "625680f99c72ef12793fea5968b6c72147fca5bf0986649c268a6a0699e0b367",
+    "trainer.pt": "2e8178c1d4313006341bdad1034631bc9d09a44086d3080d7bd8e8194fb48d2c",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
-    "performance.json": "3070ebbdf9b2b0fd6a0a237819043aa2c2a35c74a97edfa223e5c324471af4af",
-    "shard_scheduler.json": "21879a49824847fe89e795ae96e1fd2dc476cf1541cf6501892fb7519e8c9036",
-    "coverage.json": "73df3c9242608e741c41ad8c77453ef20d1273e9c95e5fe1751e37fda2234e71",
+    "performance.json": "b7cb16c0913422fe54a766da97356133a598f4da7e691939f51028a8bce75667",
+    "shard_scheduler.json": "cd21f72b325db7b3e64e3a7aac1563607bb2e1d101286d5664c18896e97ea8da",
+    "coverage.json": "35cce35db77290dc46d2183e76ef09b89428490347ea9b14cdb657a8652a8114",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 557164,
+  "step": 560000,
   "stats": {
-    "climbmix/blocks": 20057904,
-    "climbmix/targets": 33528559976,
-    "climbmix/input_tokens": 33530684873,
-    "ultra_qa/blocks": 7521714,
-    "ultra_qa/targets": 12270782181,
-    "ultra_qa/input_tokens": 12270805843,
-    "cortex/blocks": 7521714,
-    "cortex/targets": 286034884,
-    "cortex/input_tokens": 11956373662,
-    "ultra_style/blocks": 7521714,
-    "ultra_style/targets": 11896892314,
-    "ultra_style/input_tokens": 11897516222,
-    "openmath/blocks": 5014476,
-    "openmath/targets": 7748188395,
-    "openmath/input_tokens": 7750768917,
-    "rewrite6/blocks": 2507238,
-    "rewrite6/targets": 4110276401,
-    "rewrite6/input_tokens": 4114782547
+    "climbmix/blocks": 20160000,
+    "climbmix/targets": 33699449931,
+    "climbmix/input_tokens": 33701585663,
+    "ultra_qa/blocks": 7560000,
+    "ultra_qa/targets": 12333281836,
+    "ultra_qa/input_tokens": 12333305632,
+    "cortex/blocks": 7560000,
+    "cortex/targets": 287490188,
+    "cortex/input_tokens": 12017335871,
+    "ultra_style/blocks": 7560000,
+    "ultra_style/targets": 11957074205,
+    "ultra_style/input_tokens": 11957701334,
+    "openmath/blocks": 5040000,
+    "openmath/targets": 7787661032,
+    "openmath/input_tokens": 7790254729,
+    "rewrite6/blocks": 2520000,
+    "rewrite6/targets": 4131150735,
+    "rewrite6/input_tokens": 4135680080
   },
   "sources": {
     "climbmix": {
-      "page": 2747,
-      "row": 2075,
+      "page": 2761,
+      "row": 1877,
       "start": {
-        "file": 132,
+        "file": 133,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 77,
+          "row_group": 50,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 11253787
+      "offset": 4096,
+      "documents": 11310933
     },
     "rewrite6": {
-      "page": 12,
-      "row": 3016,
+      "page": 20,
+      "row": 3266,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 49152
+          "row": 81920
         }
       },
       "epoch": 31,
       "offset": 0,
-      "documents": 6419692
+      "documents": 6452710
     },
     "openmath": {
-      "page": 1716,
-      "row": 2073,
+      "page": 1725,
+      "row": 1041,
       "start": {
         "file": 16,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 42,
-          "row": 336
+          "row_group": 79,
+          "row": 200
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 7030809
+      "offset": 2048,
+      "documents": 7066641
     },
     "ultra_qa": {
-      "page": 742,
-      "row": 1162,
+      "page": 746,
+      "row": 300,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 6,
-          "row": 53994
+          "row_group": 7,
+          "row": 5982
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 3040394
+      "offset": 4096,
+      "documents": 3055916
     },
     "ultra_style": {
-      "page": 1132,
-      "row": 4042,
+      "page": 1138,
+      "row": 3363,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 6,
-          "row": 15848
+          "row": 40424
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 4640714
+      "offset": 4096,
+      "documents": 4664611
     },
     "cortex": {
-      "page": 145876,
-      "row": 97,
+      "page": 146621,
+      "row": 50,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 18672225
+      "documents": 18767538
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 557164, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "9aeb84d48b1a3e7876ab13cf2a9db637ae3eea5a8e14c759a56dfd289a32ce86", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "625680f99c72ef12793fea5968b6c72147fca5bf0986649c268a6a0699e0b367", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "3070ebbdf9b2b0fd6a0a237819043aa2c2a35c74a97edfa223e5c324471af4af", "shard_scheduler.json": "21879a49824847fe89e795ae96e1fd2dc476cf1541cf6501892fb7519e8c9036", "coverage.json": "73df3c9242608e741c41ad8c77453ef20d1273e9c95e5fe1751e37fda2234e71", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000557164"}
+{"step": 560000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "9fc6ec90f620abfbf95bf8c6ae75a133a76b595b55c08435360d181ac4591f3c", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "2e8178c1d4313006341bdad1034631bc9d09a44086d3080d7bd8e8194fb48d2c", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "b7cb16c0913422fe54a766da97356133a598f4da7e691939f51028a8bce75667", "shard_scheduler.json": "cd21f72b325db7b3e64e3a7aac1563607bb2e1d101286d5664c18896e97ea8da", "coverage.json": "35cce35db77290dc46d2183e76ef09b89428490347ea9b14cdb657a8652a8114", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000560000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9aeb84d48b1a3e7876ab13cf2a9db637ae3eea5a8e14c759a56dfd289a32ce86
+oid sha256:9fc6ec90f620abfbf95bf8c6ae75a133a76b595b55c08435360d181ac4591f3c
 size 360622744

performance.json

@@ -49,11 +49,11 @@
   "selected": {
     "microbatch": 90,
     "mode": "compiled",
-    "microbatch_seconds": 0.9447870520000379,
-    "peak_gb": 89.37766551971436,
+    "microbatch_seconds": 0.9371402589999889,
+    "peak_gb": 89.37790203094482,
     "fits_budget": true,
-    "estimated_update_seconds": 0.9447870520000379,
-    "input_tokens_per_second": 195091.58133550777
+    "estimated_update_seconds": 0.9371402589999889,
+    "input_tokens_per_second": 196683.47211620773
   },
   "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": 76613.12993076043,
-      "required_pages_per_second_upper_estimate": 0.005300274095562953
+      "required_token_positions_per_second": 76406.2405888879,
+      "required_pages_per_second_upper_estimate": 0.0052859610108428
     },
     {
       "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": 9576.641241345054,
-      "required_pages_per_second_upper_estimate": 0.004009539704741229
+      "required_token_positions_per_second": 9550.780073610988,
+      "required_pages_per_second_upper_estimate": 0.003998712173853568
     },
     {
       "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": 19153.28248269011,
-      "required_pages_per_second_upper_estimate": 0.003989123635118372
+      "required_token_positions_per_second": 19101.560147221975,
+      "required_pages_per_second_upper_estimate": 0.003978351236650072
     },
     {
       "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": 28729.923724035165,
-      "required_pages_per_second_upper_estimate": 0.0017223150592024243
+      "required_token_positions_per_second": 28652.340220832964,
+      "required_pages_per_second_upper_estimate": 0.0017176640466486026
     },
     {
       "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": 28729.923724035165,
-      "required_pages_per_second_upper_estimate": 0.002233625199964421
+      "required_token_positions_per_second": 28652.340220832964,
+      "required_pages_per_second_upper_estimate": 0.0022275934238442157
     },
     {
       "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": 28729.923724035165,
-      "required_pages_per_second_upper_estimate": 0.36851661374321987
+      "required_token_positions_per_second": 28652.340220832964,
+      "required_pages_per_second_upper_estimate": 0.36752145586681756
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,176 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 704566,
-    "producer_batch_seconds": 0.09595589186756115,
-    "producer_p95_seconds": 0.29021746900252765,
-    "gpu_update_seconds": 0.9623415733912987,
-    "producer_batches_per_second": 10.421454905345525,
-    "gpu_batches_per_second": 1.0391320791390033,
-    "producer_headroom": 10.028999310636681,
+    "producer_batch_seconds": 0.16272236082813407,
+    "producer_p95_seconds": 0.14889653499994893,
+    "gpu_update_seconds": 0.9649473581183182,
+    "producer_batches_per_second": 6.1454368957699135,
+    "gpu_batches_per_second": 1.0363259628484145,
+    "producer_headroom": 5.930023097056017,
     "sustainable": true,
-    "buffer_seconds": 60.627519123651815,
+    "buffer_seconds": 60.79168356145405,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
     "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 370915.98441075865,
-    "upload_bytes_per_second": 5961453.864962328,
+    "generation_bytes_per_second": 422396.62886957865,
+    "upload_bytes_per_second": 5499815.520343313,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 79.58993577234335,
+    "estimated_full_shards_per_hour": 90.63648366513748,
     "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": 1.272125319999759,
-      "bytes": 7866146,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.004612441000063,
-      "bytes": 5431394,
-      "pages": 26,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1393026930018095,
-      "bytes": 12031348,
-      "pages": 42,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1874672249978175,
-      "bytes": 7293811,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9514401040032681,
-      "bytes": 1189191,
-      "pages": 36,
+      "seconds": 1.022603193000009,
+      "bytes": 47630,
+      "pages": 1,
       "shards": 1
     },
     {
-      "seconds": 1.3702037240000209,
-      "bytes": 10453891,
-      "pages": 36,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9155222249974031,
-      "bytes": 2430144,
-      "pages": 31,
-      "shards": 2
-    },
-    {
-      "seconds": 1.411119290998613,
-      "bytes": 5478989,
-      "pages": 19,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9785845789992891,
-      "bytes": 1201136,
-      "pages": 33,
+      "seconds": 0.8258607079999933,
+      "bytes": 145464,
+      "pages": 4,
       "shards": 1
     },
     {
-      "seconds": 1.2566547139977047,
-      "bytes": 13157311,
-      "pages": 30,
-      "shards": 4
-    },
-    {
-      "seconds": 1.2061452220004867,
-      "bytes": 11040131,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9862021689987159,
-      "bytes": 2654656,
-      "pages": 38,
+      "seconds": 1.1176107269999989,
+      "bytes": 5541224,
+      "pages": 18,
       "shards": 2
     },
     {
-      "seconds": 0.9963805859988497,
-      "bytes": 5942893,
-      "pages": 28,
+      "seconds": 1.2054241769999408,
+      "bytes": 2385611,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.8844175519989221,
-      "bytes": 1183171,
-      "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 1.7914613989996724,
-      "bytes": 11162167,
-      "pages": 33,
-      "shards": 4
-    },
-    {
-      "seconds": 1.1609297110007901,
-      "bytes": 12171791,
-      "pages": 35,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9642307790018094,
-      "bytes": 2614472,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2208152999992308,
-      "bytes": 5885163,
-      "pages": 35,
+      "seconds": 0.8888624019999725,
+      "bytes": 6324210,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.0881164640013594,
-      "bytes": 9917140,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9187301879974257,
-      "bytes": 2590562,
-      "pages": 37,
+      "seconds": 1.1897545489999857,
+      "bytes": 1226596,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.3539272460002394,
-      "bytes": 6288510,
-      "pages": 31,
-      "shards": 2
+      "seconds": 1.2994146069999033,
+      "bytes": 17142614,
+      "pages": 34,
+      "shards": 5
     },
     {
-      "seconds": 1.2604586979978194,
-      "bytes": 6693523,
-      "pages": 28,
+      "seconds": 1.1055204159999903,
+      "bytes": 6016330,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.1304304789991875,
-      "bytes": 12005280,
-      "pages": 46,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0543503480002983,
-      "bytes": 5804640,
-      "pages": 28,
+      "seconds": 0.9128079020001678,
+      "bytes": 2402677,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.8439256270030455,
-      "bytes": 2654364,
-      "pages": 37,
+      "seconds": 1.0666157699999985,
+      "bytes": 6099291,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9141899559981539,
-      "bytes": 5548542,
-      "pages": 23,
+      "seconds": 1.0668149449998054,
+      "bytes": 5027383,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.0629937340017932,
-      "bytes": 10277378,
-      "pages": 44,
-      "shards": 3
-    },
-    {
-      "seconds": 1.4301319679980224,
-      "bytes": 8501849,
-      "pages": 41,
+      "seconds": 1.0457134279999991,
+      "bytes": 7722680,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9612259829991672,
-      "bytes": 6294173,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1198342459999822,
+      "bytes": 518226,
+      "pages": 18,
+      "shards": 1
     },
     {
-      "seconds": 1.084874779000529,
-      "bytes": 7316912,
-      "pages": 38,
+      "seconds": 1.3007597939999869,
+      "bytes": 12269409,
+      "pages": 44,
       "shards": 3
     },
     {
-      "seconds": 1.2164157670013083,
-      "bytes": 5692225,
-      "pages": 32,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0190912239995669,
-      "bytes": 6330148,
+      "seconds": 1.437565898999992,
+      "bytes": 12411381,
       "pages": 38,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0534845100009989,
-      "bytes": 2332089,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.458769362001476,
-      "bytes": 16198178,
-      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 1.0830437309996341,
-      "bytes": 7432031,
+      "seconds": 0.765915884000151,
+      "bytes": 1219645,
       "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1578626439986692,
-      "bytes": 1196629,
-      "pages": 36,
       "shards": 1
     },
     {
-      "seconds": 0.9139833829976851,
-      "bytes": 5815740,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1043080169984023,
-      "bytes": 6797860,
-      "pages": 32,
+      "seconds": 1.470795656000064,
+      "bytes": 7604342,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.0245990330004133,
-      "bytes": 6034884,
+      "seconds": 0.9404902840001341,
+      "bytes": 6153110,
       "pages": 37,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1096401040013006,
-      "bytes": 11998781,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9207302890026767,
-      "bytes": 2353345,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9312007079970499,
-      "bytes": 5937086,
-      "pages": 25,
       "shards": 2
     },
     {
-      "seconds": 0.9517514979997941,
-      "bytes": 6995458,
-      "pages": 46,
-      "shards": 3
-    },
-    {
-      "seconds": 1.261231005002628,
-      "bytes": 5881548,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9858025629982876,
-      "bytes": 6530607,
+      "seconds": 1.0696395039999516,
+      "bytes": 10812914,
       "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.5898211400017317,
-      "bytes": 12596102,
-      "pages": 36,
-      "shards": 4
-    },
-    {
-      "seconds": 0.9551758580018941,
-      "bytes": 5775284,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2868636950006476,
-      "bytes": 7532566,
-      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 0.9102224750022287,
-      "bytes": 989956,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.172385066998686,
-      "bytes": 10249964,
-      "pages": 40,
+      "seconds": 1.5107661739998548,
+      "bytes": 7696973,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0767950209992705,
-      "bytes": 2405647,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3306630199986103,
-      "bytes": 11933752,
-      "pages": 33,
-      "shards": 3
+      "seconds": 0.9512165999999525,
+      "bytes": 1125687,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 1.1605643660004716,
-      "bytes": 6187685,
+      "seconds": 1.0864234489999944,
+      "bytes": 7369290,
       "pages": 38,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1438976530007494,
-      "bytes": 6974681,
-      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.9301952720015834,
-      "bytes": 6018270,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0365287800013903,
-      "bytes": 2455252,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3521412089976366,
-      "bytes": 6109192,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.5889534400012053,
-      "bytes": 16077810,
-      "pages": 39,
-      "shards": 4
-    },
-    {
-      "seconds": 0.8900750899993,
-      "bytes": 2373140,
-      "pages": 29,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2026867530003074,
-      "bytes": 6250314,
+      "seconds": 1.1581361090002247,
+      "bytes": 10067921,
       "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0975493779988028,
-      "bytes": 5901759,
-      "pages": 32,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0364178640011232,
-      "bytes": 6898240,
-      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.1201825170028314,
-      "bytes": 6019295,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.7976953110000977,
+      "bytes": 1131427,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.523898775998532,
-      "bytes": 8240555,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.1826321499997903,
+      "bytes": 12996691,
+      "pages": 33,
+      "shards": 4
     }
   ],
   "cache_key": {
@@ -606,7 +372,7 @@
   },
   "cache_reused": true,
   "resume_probe_seconds": [
-    1.0657202160000452,
-    0.013942628999984663
+    2.6133196560000442,
+    0.09247372499999074
   ]
 }

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:625680f99c72ef12793fea5968b6c72147fca5bf0986649c268a6a0699e0b367
-size 721775807
+oid sha256:2e8178c1d4313006341bdad1034631bc9d09a44086d3080d7bd8e8194fb48d2c
+size 721753727