Qwen90M step 30000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 25000,
+  "step": 30000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "df7303e39f8a38cb32caa015de73f0c83cb91ddb60382d6c2876d432a817c840",
+    "model.safetensors": "345733703ef5ba6b63da89055fd805800d7ed30d8be589277e153476b671b19c",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "c0a2189fdee90c71c35753bc51ffdb3afecbeb4a346bd3889656ce4fe54fb477",
+    "trainer.pt": "91c1a40f9a6c8b01e9e50d09f40d34bfc35834f4940f699fc2705fad10d4fc7d",
     "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2",
     "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e",
-    "shard_scheduler.json": "61528e6c5607c5d05d8a1210d80299914e613d3ffcdb126da4557c3c47fdf788",
-    "coverage.json": "de500062e0edf2f10dd1cb2862bfcdb202710953fdc9e56a645d42f5467026c1",
+    "shard_scheduler.json": "0625ffed52cc5254b512d94fd967ab92da88751d415cb467e341553610c28e5b",
+    "coverage.json": "37ac82d062088127476b52f93c69f4abd41ed762ba2931ba3b2597dded20b2a8",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 25000,
+  "step": 30000,
   "stats": {
-    "climbmix/blocks": 900000,
-    "climbmix/targets": 1503316012,
-    "climbmix/input_tokens": 1503411635,
-    "ultra_qa/blocks": 337500,
-    "ultra_qa/targets": 550725921,
-    "ultra_qa/input_tokens": 550727011,
-    "cortex/blocks": 337500,
-    "cortex/targets": 12839020,
-    "cortex/input_tokens": 536531199,
-    "ultra_style/blocks": 337500,
-    "ultra_style/targets": 533604810,
-    "ultra_style/input_tokens": 533633023,
-    "openmath/blocks": 225000,
-    "openmath/targets": 347653521,
-    "openmath/input_tokens": 347769506,
-    "rewrite6/blocks": 112500,
-    "rewrite6/targets": 184371575,
-    "rewrite6/input_tokens": 184573987
+    "climbmix/blocks": 1080000,
+    "climbmix/targets": 1804038329,
+    "climbmix/input_tokens": 1804152840,
+    "ultra_qa/blocks": 405000,
+    "ultra_qa/targets": 660784533,
+    "ultra_qa/input_tokens": 660785848,
+    "cortex/blocks": 405000,
+    "cortex/targets": 15406772,
+    "cortex/input_tokens": 643853720,
+    "ultra_style/blocks": 405000,
+    "ultra_style/targets": 640617580,
+    "ultra_style/input_tokens": 640651016,
+    "openmath/blocks": 270000,
+    "openmath/targets": 417204122,
+    "openmath/input_tokens": 417343326,
+    "rewrite6/blocks": 135000,
+    "rewrite6/targets": 221241868,
+    "rewrite6/input_tokens": 221484805
   },
   "sources": {
     "climbmix": {
-      "page": 123,
-      "row": 2056,
+      "page": 148,
+      "row": 423,
       "start": {
-        "file": 5,
+        "file": 7,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 76,
+          "row_group": 11,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 505864
+      "documents": 606631
     },
     "rewrite6": {
-      "page": 20,
-      "row": 960,
+      "page": 34,
+      "row": 1394,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 81920
+          "row": 139264
         }
       },
       "epoch": 1,
-      "offset": 2048,
-      "documents": 288284
+      "offset": 0,
+      "documents": 346062
     },
     "openmath": {
-      "page": 77,
-      "row": 320,
+      "page": 92,
+      "row": 2045,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 315,
-          "row": 392
+          "row_group": 376,
+          "row": 832
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 315712
+      "documents": 378877
     },
     "ultra_qa": {
-      "page": 33,
-      "row": 1362,
+      "page": 40,
+      "row": 85,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 2,
-          "row": 5892
+          "row": 34564
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 136530
+      "documents": 163925
     },
     "ultra_style": {
-      "page": 50,
-      "row": 3964,
+      "page": 60,
+      "row": 3939,
       "start": {
         "file": 0,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 2,
-          "row": 33851
+          "row": 74811
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 208764
+      "documents": 249699
     },
     "cortex": {
-      "page": 6544,
-      "row": 9,
+      "page": 7853,
+      "row": 114,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 837641
+      "documents": 1005298
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 25000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "df7303e39f8a38cb32caa015de73f0c83cb91ddb60382d6c2876d432a817c840", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "c0a2189fdee90c71c35753bc51ffdb3afecbeb4a346bd3889656ce4fe54fb477", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "61528e6c5607c5d05d8a1210d80299914e613d3ffcdb126da4557c3c47fdf788", "coverage.json": "de500062e0edf2f10dd1cb2862bfcdb202710953fdc9e56a645d42f5467026c1", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000025000"}
+{"step": 30000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "345733703ef5ba6b63da89055fd805800d7ed30d8be589277e153476b671b19c", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "91c1a40f9a6c8b01e9e50d09f40d34bfc35834f4940f699fc2705fad10d4fc7d", "training_config.json": "07f331dbc4eba0ea92b45763c59c20af4874302e74a327f3562642d69645fed2", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "0625ffed52cc5254b512d94fd967ab92da88751d415cb467e341553610c28e5b", "coverage.json": "37ac82d062088127476b52f93c69f4abd41ed762ba2931ba3b2597dded20b2a8", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000030000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:df7303e39f8a38cb32caa015de73f0c83cb91ddb60382d6c2876d432a817c840
+oid sha256:345733703ef5ba6b63da89055fd805800d7ed30d8be589277e153476b671b19c
 size 360622744

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": 75999.12812604562,
-      "required_pages_per_second_upper_estimate": 0.006587411807936398
+      "required_token_positions_per_second": 75157.17042322725,
+      "required_pages_per_second_upper_estimate": 0.00651443304817839
     },
     {
       "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": 9499.891015755702,
-      "required_pages_per_second_upper_estimate": 0.0037896485622130613
+      "required_token_positions_per_second": 9394.646302903406,
+      "required_pages_per_second_upper_estimate": 0.0037476648727075975
     },
     {
       "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": 18999.782031511404,
-      "required_pages_per_second_upper_estimate": 0.004395265188247066
+      "required_token_positions_per_second": 18789.29260580681,
+      "required_pages_per_second_upper_estimate": 0.004346572164097676
     },
     {
       "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": 28499.673047267104,
-      "required_pages_per_second_upper_estimate": 0.0017354805623500836
+      "required_token_positions_per_second": 28183.938908710217,
+      "required_pages_per_second_upper_estimate": 0.0017162540098409707
     },
     {
       "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": 28499.673047267104,
-      "required_pages_per_second_upper_estimate": 0.0021939837233229385
+      "required_token_positions_per_second": 28183.938908710217,
+      "required_pages_per_second_upper_estimate": 0.002169677635328795
     },
     {
       "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": 28499.673047267104,
-      "required_pages_per_second_upper_estimate": 0.3538705569771298
+      "required_token_positions_per_second": 28183.938908710217,
+      "required_pages_per_second_upper_estimate": 0.3499501956704399
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,409 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.12395837693742351,
-    "producer_p95_seconds": 0.32236783699772786,
-    "gpu_update_seconds": 0.970116392358095,
-    "producer_batches_per_second": 8.067224052996584,
-    "gpu_batches_per_second": 1.0308041466748807,
-    "producer_headroom": 7.826146294637495,
+    "producer_batch_seconds": 0.3413639609687493,
+    "producer_p95_seconds": 0.3689845970002352,
+    "gpu_update_seconds": 0.9809842438827957,
+    "producer_batches_per_second": 2.929424644482452,
+    "gpu_batches_per_second": 1.0193843644643452,
+    "producer_headroom": 2.8737194198792455,
     "sustainable": true,
-    "buffer_seconds": 60.14721632620189,
+    "buffer_seconds": 60.82102312073334,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 80,
+    "upload_pages_per_commit": 82,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 466257.7871084776,
-    "upload_bytes_per_second": 5728821.4302538335,
+    "generation_bytes_per_second": 341581.6420395341,
+    "upload_bytes_per_second": 6088898.701746191,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 100.04806718769785,
+    "estimated_full_shards_per_hour": 73.29546876801984,
     "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": 1.0224713220013655,
-      "bytes": 6909652,
-      "pages": 35,
-      "shards": 2
+      "seconds": 0.808165985999949,
+      "bytes": 1135299,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.018912650000857,
-      "bytes": 5819747,
-      "pages": 22,
+      "seconds": 1.9975283729982038,
+      "bytes": 6018894,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.8826371689974621,
-      "bytes": 4172226,
-      "pages": 46,
+      "seconds": 1.1784233340003993,
+      "bytes": 7683890,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 2.508591282999987,
-      "bytes": 5665007,
-      "pages": 21,
-      "shards": 2
+      "seconds": 1.1247381559987844,
+      "bytes": 11305256,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.1267430049992981,
-      "bytes": 4735418,
-      "pages": 32,
-      "shards": 2
+      "seconds": 0.900414362000447,
+      "bytes": 7293854,
+      "pages": 29,
+      "shards": 3
     },
     {
-      "seconds": 1.0588101130015275,
-      "bytes": 8209294,
-      "pages": 47,
-      "shards": 4
+      "seconds": 1.0943440850023762,
+      "bytes": 1102979,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 1.0802435129990045,
-      "bytes": 11583568,
-      "pages": 29,
+      "seconds": 1.0738140280009247,
+      "bytes": 6022869,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.2433872929977952,
-      "bytes": 7696542,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.1826581230016018,
+      "bytes": 10993511,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 0.9942282960000739,
-      "bytes": 5730133,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.020132988000114,
+      "bytes": 882161,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 0.9958830260002287,
-      "bytes": 6097937,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.030573132000427,
+      "bytes": 11789498,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.0204584569983126,
-      "bytes": 2643615,
+      "seconds": 0.8314693719985371,
+      "bytes": 2720719,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1489640459985822,
-      "bytes": 6010015,
-      "pages": 32,
+      "seconds": 0.9525506049976684,
+      "bytes": 10860039,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.2853242279998085,
+      "bytes": 7445426,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 0.7597367620001023,
+      "bytes": 831447,
+      "pages": 25,
+      "shards": 1
+    },
+    {
+      "seconds": 1.0118765539991728,
+      "bytes": 6408285,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3807983080005215,
-      "bytes": 15814135,
-      "pages": 39,
+      "seconds": 1.0685828210007458,
+      "bytes": 12919611,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 1.1238995200001227,
-      "bytes": 2339467,
+      "seconds": 1.0399747580013354,
+      "bytes": 5973768,
       "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 1.365642591001233,
-      "bytes": 6320146,
+      "seconds": 1.0144561509987398,
+      "bytes": 6361193,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0016697349965398,
-      "bytes": 761758,
-      "pages": 22,
+      "seconds": 1.0571298460017715,
+      "bytes": 2527314,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1483951949994662,
-      "bytes": 11735955,
-      "pages": 47,
-      "shards": 4
+      "seconds": 1.1346876719981083,
+      "bytes": 10084333,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.0657177410030272,
-      "bytes": 1129396,
-      "pages": 33,
-      "shards": 1
+      "seconds": 0.9850222569984908,
+      "bytes": 2356425,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.0460155590008071,
-      "bytes": 6970566,
-      "pages": 24,
+      "seconds": 1.0524133049984812,
+      "bytes": 12178249,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 2.029252282998641,
-      "bytes": 16530913,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.109467931000836,
+      "bytes": 5783909,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.8704905230006261,
-      "bytes": 1186552,
-      "pages": 33,
-      "shards": 1
+      "seconds": 0.8641120300017064,
+      "bytes": 2329685,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 3.2236978209984954,
-      "bytes": 7889017,
-      "pages": 38,
+      "seconds": 1.4142594379991351,
+      "bytes": 10485303,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 0.9988030870008515,
-      "bytes": 1104863,
-      "pages": 33,
-      "shards": 1
+      "seconds": 0.9442589440004667,
+      "bytes": 6417253,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.0168222699976468,
-      "bytes": 4937209,
-      "pages": 2,
+      "seconds": 0.9173370479984442,
+      "bytes": 2397470,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0983402380006737,
-      "bytes": 4472235,
-      "pages": 20,
-      "shards": 2
+      "seconds": 1.938626930997998,
+      "bytes": 16310135,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.3725059809985396,
-      "bytes": 7833557,
-      "pages": 41,
-      "shards": 3
+      "seconds": 1.1073444919966278,
+      "bytes": 2450943,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.04416848999972,
-      "bytes": 7007035,
-      "pages": 42,
+      "seconds": 0.8343555079991347,
+      "bytes": 5999582,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.018632732997503,
-      "bytes": 7180638,
-      "pages": 29,
+      "seconds": 0.9876751570009219,
+      "bytes": 1005909,
+      "pages": 31,
+      "shards": 1
+    },
+    {
+      "seconds": 1.0930767579993699,
+      "bytes": 7383476,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.1994012059985835,
-      "bytes": 10151059,
-      "pages": 38,
+      "seconds": 1.2710551140007738,
+      "bytes": 11134870,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 0.9821173639975314,
-      "bytes": 2386369,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.053878970000369,
+      "bytes": 8139088,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.2013670170017576,
-      "bytes": 6446691,
-      "pages": 35,
+      "seconds": 1.03730981300032,
+      "bytes": 6215262,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0164053049993527,
-      "bytes": 1100982,
-      "pages": 30,
+      "seconds": 0.9270265450031729,
+      "bytes": 1100519,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.2673739959973318,
-      "bytes": 17927773,
-      "pages": 41,
-      "shards": 5
+      "seconds": 1.1968161780023365,
+      "bytes": 11473828,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 0.8297740179987159,
-      "bytes": 928094,
+      "seconds": 0.9470581830028095,
+      "bytes": 5899775,
       "pages": 28,
-      "shards": 1
+      "shards": 3
     },
     {
-      "seconds": 0.8634128480007348,
-      "bytes": 6113242,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.0518160799983889,
+      "bytes": 1140036,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 0.979562092001288,
-      "bytes": 7577775,
-      "pages": 38,
+      "seconds": 1.1662064560005092,
+      "bytes": 7349596,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 0.9862535339998431,
-      "bytes": 4746157,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.122332668001036,
+      "bytes": 11201873,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.0129832909988181,
-      "bytes": 7553178,
+      "seconds": 0.8321514729977935,
+      "bytes": 6044525,
       "pages": 36,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.2109974509985477,
-      "bytes": 11754963,
-      "pages": 35,
+      "seconds": 1.4008962299994892,
+      "bytes": 7557731,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.878410489000089,
-      "bytes": 971318,
-      "pages": 31,
+      "seconds": 0.8883114669988572,
+      "bytes": 823542,
+      "pages": 25,
       "shards": 1
     },
     {
-      "seconds": 1.079362858999957,
-      "bytes": 12108906,
-      "pages": 37,
+      "seconds": 1.2165515029992093,
+      "bytes": 11904873,
+      "pages": 42,
       "shards": 4
     },
     {
-      "seconds": 0.808165985999949,
-      "bytes": 1135299,
-      "pages": 34,
+      "seconds": 0.9436809830003767,
+      "bytes": 889392,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.9975283729982038,
-      "bytes": 6018894,
-      "pages": 27,
-      "shards": 2
+      "seconds": 1.1113080279974383,
+      "bytes": 11835067,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.1784233340003993,
-      "bytes": 7683890,
+      "seconds": 0.9863674269981857,
+      "bytes": 7525026,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1247381559987844,
-      "bytes": 11305256,
-      "pages": 34,
-      "shards": 3
+      "seconds": 0.7792704010025773,
+      "bytes": 5012728,
+      "pages": 27,
+      "shards": 2
     },
     {
-      "seconds": 0.900414362000447,
-      "bytes": 7293854,
-      "pages": 29,
+      "seconds": 1.0204439990011451,
+      "bytes": 7601581,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0943440850023762,
-      "bytes": 1102979,
+      "seconds": 0.844439466996846,
+      "bytes": 1090867,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.0738140280009247,
-      "bytes": 6022869,
-      "pages": 36,
-      "shards": 3
+      "seconds": 0.8992618750016845,
+      "bytes": 5707571,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.1826581230016018,
-      "bytes": 10993511,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.1450656299966795,
+      "bytes": 17577071,
+      "pages": 37,
+      "shards": 5
     },
     {
-      "seconds": 1.020132988000114,
-      "bytes": 882161,
+      "seconds": 1.0943805140013865,
+      "bytes": 891607,
       "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.030573132000427,
-      "bytes": 11789498,
-      "pages": 32,
-      "shards": 3
-    },
-    {
-      "seconds": 0.8314693719985371,
-      "bytes": 2720719,
-      "pages": 36,
+      "seconds": 0.9035562749995734,
+      "bytes": 6454400,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9525506049976684,
-      "bytes": 10860039,
-      "pages": 34,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2853242279998085,
-      "bytes": 7445426,
+      "seconds": 1.2266940429981332,
+      "bytes": 2571680,
       "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 0.7597367620001023,
-      "bytes": 831447,
-      "pages": 25,
-      "shards": 1
-    },
-    {
-      "seconds": 1.0118765539991728,
-      "bytes": 6408285,
-      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0685828210007458,
-      "bytes": 12919611,
-      "pages": 37,
-      "shards": 4
+      "seconds": 1.0315104050023365,
+      "bytes": 10269533,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.0399747580013354,
-      "bytes": 5973768,
-      "pages": 27,
+      "seconds": 0.8940106660011224,
+      "bytes": 5798800,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.0144561509987398,
-      "bytes": 6361193,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.0886239710016525,
+      "bytes": 8189805,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.0571298460017715,
-      "bytes": 2527314,
+      "seconds": 0.7768403989975923,
+      "bytes": 6370933,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1346876719981083,
-      "bytes": 10084333,
-      "pages": 34,
+      "seconds": 1.1446712159995514,
+      "bytes": 7008906,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 0.9850222569984908,
-      "bytes": 2356425,
-      "pages": 28,
+      "seconds": 0.9077771720003511,
+      "bytes": 6012024,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0524133049984812,
-      "bytes": 12178249,
-      "pages": 40,
-      "shards": 3
+      "seconds": 1.4875493239997013,
+      "bytes": 6032602,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.109467931000836,
-      "bytes": 5783909,
-      "pages": 28,
+      "seconds": 0.8995478449978691,
+      "bytes": 2436827,
+      "pages": 30,
       "shards": 2
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c0a2189fdee90c71c35753bc51ffdb3afecbeb4a346bd3889656ce4fe54fb477
-size 721767167
+oid sha256:91c1a40f9a6c8b01e9e50d09f40d34bfc35834f4940f699fc2705fad10d4fc7d
+size 721767935