Qwen90M step 45000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 40000,
+  "step": 45000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "1f54edcf173e14eaa220773d6904ca3660dca9333bec4cca80621f41c11ae4ba",
+    "model.safetensors": "8538b491fbf01a73501462b1b8b4d11ed8192e49c9b77a3ed172e5e05d009d88",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "ddb86ed5b8deb7d4e788e14749c412fa48be7f3cefd33dadef82d54b10488b02",
-    "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2",
-    "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e",
-    "shard_scheduler.json": "dd35db67300912a92de8d12b39d40858b1385a9f601def538c82db10aecb48b5",
-    "coverage.json": "496553c7362fb891e4a087907a1ee7bc57bd29676eca9ccbf872f730f0c086c9",
+    "trainer.pt": "d5d9f25a02093a0b4efb7c3de3f9c0397ade770ae17612d84206b659abe4a816",
+    "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
+    "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f",
+    "shard_scheduler.json": "1fa5611cb9e27a2e65056abc0c9c52bc5694633d8adb64378769c8274e989bc6",
+    "coverage.json": "e0f2cf7ac209336f996ae01d22aab46c0dbcfe6c255e055d3cc863bf5d710ea4",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 40000,
+  "step": 45000,
   "stats": {
-    "climbmix/blocks": 1440000,
-    "climbmix/targets": 2405645996,
-    "climbmix/input_tokens": 2405798466,
-    "ultra_qa/blocks": 540000,
-    "ultra_qa/targets": 881103422,
-    "ultra_qa/input_tokens": 881105161,
-    "cortex/blocks": 540000,
-    "cortex/targets": 20545868,
-    "cortex/input_tokens": 858494555,
-    "ultra_style/blocks": 540000,
-    "ultra_style/targets": 853948670,
-    "ultra_style/input_tokens": 853993129,
-    "openmath/blocks": 360000,
-    "openmath/targets": 556374754,
-    "openmath/input_tokens": 556560449,
-    "rewrite6/blocks": 180000,
-    "rewrite6/targets": 295083056,
-    "rewrite6/input_tokens": 295406501
+    "climbmix/blocks": 1620000,
+    "climbmix/targets": 2706557052,
+    "climbmix/input_tokens": 2706728542,
+    "ultra_qa/blocks": 607500,
+    "ultra_qa/targets": 991206075,
+    "ultra_qa/input_tokens": 991208036,
+    "cortex/blocks": 607500,
+    "cortex/targets": 23108530,
+    "cortex/input_tokens": 965718686,
+    "ultra_style/blocks": 607500,
+    "ultra_style/targets": 960329583,
+    "ultra_style/input_tokens": 960379775,
+    "openmath/blocks": 405000,
+    "openmath/targets": 625892819,
+    "openmath/input_tokens": 626101608,
+    "rewrite6/blocks": 202500,
+    "rewrite6/targets": 331937575,
+    "rewrite6/input_tokens": 332301831
   },
   "sources": {
     "climbmix": {
-      "page": 197,
-      "row": 1098,
+      "page": 221,
+      "row": 3842,
       "start": {
-        "file": 9,
+        "file": 10,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 43,
+          "row_group": 56,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 808010
+      "offset": 2048,
+      "documents": 909058
     },
     "rewrite6": {
-      "page": 12,
-      "row": 807,
+      "page": 26,
+      "row": 1558,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 49152
+          "row": 106496
         }
       },
       "epoch": 2,
       "offset": 0,
-      "documents": 460767
+      "documents": 518862
     },
     "openmath": {
-      "page": 123,
-      "row": 1391,
+      "page": 138,
+      "row": 2902,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 67,
-          "row": 158
+          "row_group": 128,
+          "row": 598
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 505199
+      "documents": 568150
     },
     "ultra_qa": {
-      "page": 53,
-      "row": 1357,
+      "page": 60,
+      "row": 12,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 3,
-          "row": 23181
+          "row": 51853
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 218445
+      "documents": 245772
     },
     "ultra_style": {
-      "page": 81,
-      "row": 1172,
+      "page": 91,
+      "row": 2493,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 3,
-          "row": 76752
+          "row_group": 4,
+          "row": 32715
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 332948
+      "offset": 2048,
+      "documents": 375229
     },
     "cortex": {
-      "page": 10471,
-      "row": 124,
+      "page": 11778,
+      "row": 67,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 1340412
+      "documents": 1507651
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 40000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "1f54edcf173e14eaa220773d6904ca3660dca9333bec4cca80621f41c11ae4ba", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "ddb86ed5b8deb7d4e788e14749c412fa48be7f3cefd33dadef82d54b10488b02", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "dd35db67300912a92de8d12b39d40858b1385a9f601def538c82db10aecb48b5", "coverage.json": "496553c7362fb891e4a087907a1ee7bc57bd29676eca9ccbf872f730f0c086c9", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000040000"}
+{"step": 45000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "8538b491fbf01a73501462b1b8b4d11ed8192e49c9b77a3ed172e5e05d009d88", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "d5d9f25a02093a0b4efb7c3de3f9c0397ade770ae17612d84206b659abe4a816", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "d14cb208816158eb276e72d732c3b9e6358b2a75509678412180fc272350709f", "shard_scheduler.json": "1fa5611cb9e27a2e65056abc0c9c52bc5694633d8adb64378769c8274e989bc6", "coverage.json": "e0f2cf7ac209336f996ae01d22aab46c0dbcfe6c255e055d3cc863bf5d710ea4", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000045000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1f54edcf173e14eaa220773d6904ca3660dca9333bec4cca80621f41c11ae4ba
+oid sha256:8538b491fbf01a73501462b1b8b4d11ed8192e49c9b77a3ed172e5e05d009d88
 size 360622744

performance.json

@@ -44,16 +44,16 @@
     "source": "e6d135a4487a9fd1907e12903cda79e75fd61305fad149fda7ec5281279dc18b",
     "requested_microbatch_ceiling": 90
   },
-  "cache_reused": false,
+  "cache_reused": true,
   "attention": "PyTorch FlashAttention (forced; forward/backward verified)",
   "selected": {
     "microbatch": 86,
     "mode": "compiled",
-    "microbatch_seconds": 0.8818629400000191,
-    "peak_gb": 84.78368663787842,
+    "microbatch_seconds": 0.8974938420000171,
+    "peak_gb": 85.39913082122803,
     "fits_budget": true,
-    "estimated_update_seconds": 0.9349941874999956,
-    "input_tokens_per_second": 199722.64624250587
+    "estimated_update_seconds": 0.9533413040000482,
+    "input_tokens_per_second": 196244.2434228943
   },
   "measurements": [
     {
@@ -218,5 +218,6 @@
   "selection_policy": "max_batch",
   "budget_gb": 89.6964394569397,
   "reserve_gb": 2.1791439056396484,
-  "note": "Forward/backward throughput; excludes optimizer, data, logging and checkpoint time."
+  "note": "Forward/backward throughput; excludes optimizer, data, logging and checkpoint time.",
+  "legacy_seed_verified": false
 }

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": 75414.82490767069,
-      "required_pages_per_second_upper_estimate": 0.006536765885870627
+      "required_token_positions_per_second": 75734.50043516197,
+      "required_pages_per_second_upper_estimate": 0.006564474550383355
     },
     {
       "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": 9426.853113458836,
-      "required_pages_per_second_upper_estimate": 0.003760512650972888
+      "required_token_positions_per_second": 9466.812554395246,
+      "required_pages_per_second_upper_estimate": 0.0037764530694093055
     },
     {
       "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": 18853.70622691767,
-      "required_pages_per_second_upper_estimate": 0.004361473121700661
+      "required_token_positions_per_second": 18933.625108790493,
+      "required_pages_per_second_upper_estimate": 0.004379960948497657
     },
     {
       "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": 28280.55934037651,
-      "required_pages_per_second_upper_estimate": 0.0017221376872012205
+      "required_token_positions_per_second": 28400.437663185738,
+      "required_pages_per_second_upper_estimate": 0.0017294376481073507
     },
     {
       "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": 28280.55934037651,
-      "required_pages_per_second_upper_estimate": 0.002177115743620939
+      "required_token_positions_per_second": 28400.437663185738,
+      "required_pages_per_second_upper_estimate": 0.00218634430875523
     },
     {
       "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": 28280.55934037651,
-      "required_pages_per_second_upper_estimate": 0.3511498980639521
+      "required_token_positions_per_second": 28400.437663185738,
+      "required_pages_per_second_upper_estimate": 0.3526383856263052
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,140 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.1152253453594767,
-    "producer_p95_seconds": 0.4034826810020604,
-    "gpu_update_seconds": 0.9776327146587446,
-    "producer_batches_per_second": 8.678646150985522,
-    "gpu_batches_per_second": 1.0228790270680161,
-    "producer_headroom": 8.48452839615064,
+    "producer_batch_seconds": 0.053367566671880695,
+    "producer_p95_seconds": 0.17484046899971872,
+    "gpu_update_seconds": 0.9735061243735307,
+    "producer_batches_per_second": 18.737972562029867,
+    "gpu_batches_per_second": 1.0272149039057341,
+    "producer_headroom": 18.241531047479253,
     "sustainable": true,
-    "buffer_seconds": 60.61322830884216,
+    "buffer_seconds": 60.3573797111589,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 346780.9633111328,
-    "upload_bytes_per_second": 3718135.8829697864,
+    "generation_bytes_per_second": 449162.1108758026,
+    "upload_bytes_per_second": 6093636.035551512,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 74.41112207890022,
+    "estimated_full_shards_per_hour": 96.37973303514059,
     "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": 0.9655926410050597,
-      "bytes": 5648372,
-      "pages": 25,
-      "shards": 2
-    },
-    {
-      "seconds": 2.317049978002615,
-      "bytes": 7094864,
-      "pages": 41,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0156349210010376,
-      "bytes": 12015881,
-      "pages": 35,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0830463239981327,
-      "bytes": 6195752,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 0.7641328610043274,
-      "bytes": 2288843,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.014886224998918,
-      "bytes": 6118927,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.0632401930001834,
+      "bytes": 22792,
+      "pages": 1,
+      "shards": 1
     },
     {
-      "seconds": 0.9393350060054217,
-      "bytes": 807847,
-      "pages": 26,
+      "seconds": 0.9811290180000469,
+      "bytes": 1073130,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.4682032510027057,
-      "bytes": 16949786,
-      "pages": 39,
+      "seconds": 1.3945427510002446,
+      "bytes": 16977562,
+      "pages": 43,
       "shards": 5
     },
     {
-      "seconds": 0.9623997290036641,
-      "bytes": 6092433,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 0.7187483379966579,
-      "bytes": 1132933,
-      "pages": 34,
+      "seconds": 0.7539118750000853,
+      "bytes": 789098,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 1.4302320659990073,
-      "bytes": 11751486,
+      "seconds": 0.9344837560001906,
+      "bytes": 6249982,
       "pages": 35,
-      "shards": 4
-    },
-    {
-      "seconds": 0.8936330840006121,
-      "bytes": 843076,
-      "pages": 27,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0585707080026623,
-      "bytes": 7490877,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 33.798858237001696,
-      "bytes": 12238534,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 0.911944024002878,
-      "bytes": 5136667,
-      "pages": 41,
       "shards": 2
     },
     {
-      "seconds": 1.2792566019998048,
-      "bytes": 7551543,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 0.9091981460005627,
-      "bytes": 5814772,
-      "pages": 27,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0604712630010908,
-      "bytes": 6743320,
-      "pages": 36,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0128854309950839,
-      "bytes": 6059900,
-      "pages": 26,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0482589550010744,
-      "bytes": 7056720,
-      "pages": 42,
-      "shards": 2
-    },
-    {
-      "seconds": 1.9086563699966064,
-      "bytes": 7678172,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1078665649984032,
-      "bytes": 9761910,
-      "pages": 33,
-      "shards": 3
-    },
-    {
-      "seconds": 0.90664235100121,
-      "bytes": 2695308,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 0.7943546639944543,
-      "bytes": 258770,
-      "pages": 8,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1189483790003578,
-      "bytes": 5921688,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2605290909996256,
-      "bytes": 6824656,
-      "pages": 33,
-      "shards": 2
-    },
-    {
-      "seconds": 11.391802897000161,
-      "bytes": 12565360,
-      "pages": 37,
-      "shards": 4
-    },
-    {
-      "seconds": 0.941881225000543,
-      "bytes": 6023516,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9670609070017235,
-      "bytes": 2689085,
+      "seconds": 1.0289287979999244,
+      "bytes": 7839831,
       "pages": 35,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1650014270053362,
-      "bytes": 5886860,
-      "pages": 31,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1598776869941503,
-      "bytes": 10363400,
-      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1336681269967812,
-      "bytes": 8324153,
+      "seconds": 1.02437976300007,
+      "bytes": 6057442,
       "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0718341040046653,
-      "bytes": 6483305,
-      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9573828110005707,
-      "bytes": 873310,
-      "pages": 27,
-      "shards": 1
-    },
-    {
-      "seconds": 1.2830225369980326,
-      "bytes": 12398676,
-      "pages": 40,
-      "shards": 4
-    },
-    {
-      "seconds": 1.2898951360039064,
-      "bytes": 6161152,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9979428110018489,
-      "bytes": 2470155,
-      "pages": 30,
-      "shards": 2
-    },
-    {
-      "seconds": 1.098158920998685,
-      "bytes": 11654949,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0713990720032598,
-      "bytes": 5874732,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9352636539988453,
-      "bytes": 6076485,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2692730029957602,
-      "bytes": 6086693,
+      "seconds": 1.1024301279999236,
+      "bytes": 7712415,
       "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.036868928000331,
-      "bytes": 2682212,
-      "pages": 38,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1587374009977793,
-      "bytes": 10973458,
-      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 0.8646178900016821,
-      "bytes": 947338,
-      "pages": 27,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0228583920033998,
-      "bytes": 13452885,
-      "pages": 43,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0817572929954622,
-      "bytes": 6130820,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9214774369975203,
-      "bytes": 1243052,
+      "seconds": 0.9960975080002754,
+      "bytes": 6877408,
       "pages": 35,
-      "shards": 1
-    },
-    {
-      "seconds": 1.3106369909946807,
-      "bytes": 11452061,
-      "pages": 33,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0961634519990184,
-      "bytes": 6066738,
-      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9469924939985503,
-      "bytes": 890694,
-      "pages": 27,
+      "seconds": 1.0195401410001068,
+      "bytes": 6014632,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.2476149669964798,
-      "bytes": 13021486,
-      "pages": 30,
+      "seconds": 1.221793560000151,
+      "bytes": 10970126,
+      "pages": 34,
       "shards": 4
     },
     {
-      "seconds": 0.8867707609970239,
-      "bytes": 6143048,
-      "pages": 43,
-      "shards": 2
-    },
-    {
-      "seconds": 1.05023945299763,
-      "bytes": 7400872,
-      "pages": 31,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1047442850031075,
-      "bytes": 6203278,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0125872680000612,
-      "bytes": 1072664,
-      "pages": 31,
+      "seconds": 0.999679897000533,
+      "bytes": 1109612,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.132626735001395,
-      "bytes": 11459123,
-      "pages": 36,
-      "shards": 4
-    },
-    {
-      "seconds": 1.060930259001907,
-      "bytes": 6808205,
-      "pages": 34,
+      "seconds": 0.9920720150003035,
+      "bytes": 6122911,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 3.3303450339953997,
-      "bytes": 5952538,
+      "seconds": 1.0328473609997673,
+      "bytes": 7480746,
       "pages": 29,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 0.9529956519982079,
-      "bytes": 7196338,
-      "pages": 30,
+      "seconds": 1.141071416000159,
+      "bytes": 10868715,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.1464728400023887,
-      "bytes": 6003121,
+      "seconds": 1.1746170310007074,
+      "bytes": 7476573,
       "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9798986519963364,
-      "bytes": 7781027,
-      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9489700769991032,
-      "bytes": 911408,
-      "pages": 26,
+      "seconds": 1.0039610859994355,
+      "bytes": 1092998,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.3251626939963899,
-      "bytes": 11870118,
-      "pages": 36,
+      "seconds": 1.1609742079999705,
+      "bytes": 10556650,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.1823667450007633,
-      "bytes": 10876587,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.1113881370001764,
+      "bytes": 7415466,
+      "pages": 37,
+      "shards": 3
     }
   ],
   "cache_key": {
@@ -604,5 +334,9 @@
     "cpu_count": 20,
     "microbatch": 86
   },
-  "cache_reused": false
+  "cache_reused": true,
+  "resume_probe_seconds": [
+    2.657314213999996,
+    0.016167582000036873
+  ]
 }

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ddb86ed5b8deb7d4e788e14749c412fa48be7f3cefd33dadef82d54b10488b02
-size 721776959
+oid sha256:d5d9f25a02093a0b4efb7c3de3f9c0397ade770ae17612d84206b659abe4a816
+size 721769471

training_config.json

@@ -71,5 +71,7 @@
   "hub_resume": true,
   "work_dir": "/marimo/qwen90m_workspace/runs/void-byte-v1",
   "data_recipe": "c0b0679d54277afcf4c85726fd0998265697a759188400782be4a859641ba658",
-  "trainer_recipe": "bc45094ca6fe6129cd1201f38cc5978b80014fa21be8d01ea684cb56d8d1644b"
+  "trainer_recipe": "bc45094ca6fe6129cd1201f38cc5978b80014fa21be8d01ea684cb56d8d1644b",
+  "archive_recipe": "c0b0679d54277afcf4c85726fd0998265697a759188400782be4a859641ba658",
+  "cortex_archive_recipe": "c0b0679d54277afcf4c85726fd0998265697a759188400782be4a859641ba658"
 }