Qwen90M step 645000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 640000,
+  "step": 645000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "ebc22606ea875e024664b2b5ba82033f919f8d1a96f22dfebda33a7cb5296c83",
+    "model.safetensors": "63dedfe6f429aac5ad00d42078291bc517b71b9421a80358aeab37e9ea3770b0",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "2e88f84323ca4bdc4a8a0fa1302e353f3a442dcb7967001b5b87a164915124f0",
+    "trainer.pt": "09595760b7624b74f118f3a6bb0623c52c6fe05ffe26da58b1c6e977fdb298ab",
     "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb",
     "performance.json": "539995c1e4203016bfc6259363445c7c25dfb51604d74cd18d2cc7ad51bfb71e",
-    "shard_scheduler.json": "bf475a77a1c1a0c722213e8028f3d9758aacd623f550f0da5d55fe94219c3d81",
-    "coverage.json": "33bc21ae73a3caaa60d3061ffb19116a58b5eaf2262d82489a7714a0458299ba",
+    "shard_scheduler.json": "e3524f5d2e0a02ee5df1bd332f50fbbc4165917e20769e839cc84064c0934bfc",
+    "coverage.json": "63498e908791e3993c2423818a474742357469ed5acc8033f03f8f9c55147e00",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 640000,
+  "step": 645000,
   "stats": {
-    "climbmix/blocks": 23040000,
-    "climbmix/targets": 38516782526,
-    "climbmix/input_tokens": 38519223105,
-    "ultra_qa/blocks": 8640000,
-    "ultra_qa/targets": 14095882783,
-    "ultra_qa/input_tokens": 14095910016,
-    "cortex/blocks": 8640000,
-    "cortex/targets": 328530053,
-    "cortex/input_tokens": 13733804797,
-    "ultra_style/blocks": 8640000,
-    "ultra_style/targets": 13665830277,
-    "ultra_style/input_tokens": 13666547133,
-    "openmath/blocks": 5760000,
-    "openmath/targets": 8900224915,
-    "openmath/input_tokens": 8903189812,
-    "rewrite6/blocks": 2880000,
-    "rewrite6/targets": 4721409048,
-    "rewrite6/input_tokens": 4726585429
+    "climbmix/blocks": 23220000,
+    "climbmix/targets": 38817162199,
+    "climbmix/input_tokens": 38819622039,
+    "ultra_qa/blocks": 8707500,
+    "ultra_qa/targets": 14205974024,
+    "ultra_qa/input_tokens": 14206001475,
+    "cortex/blocks": 8707500,
+    "cortex/targets": 331096706,
+    "cortex/input_tokens": 13841071788,
+    "ultra_style/blocks": 8707500,
+    "ultra_style/targets": 13772861168,
+    "ultra_style/input_tokens": 13773583331,
+    "openmath/blocks": 5805000,
+    "openmath/targets": 8969780734,
+    "openmath/input_tokens": 8972768822,
+    "rewrite6/blocks": 2902500,
+    "rewrite6/targets": 4758276221,
+    "rewrite6/input_tokens": 4763492845
   },
   "sources": {
     "climbmix": {
-      "page": 3155,
-      "row": 394,
+      "page": 3179,
+      "row": 3644,
       "start": {
-        "file": 152,
+        "file": 153,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 60,
+          "row_group": 73,
           "row": 1024
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 12923274
+      "documents": 13024828
     },
     "rewrite6": {
-      "page": 45,
-      "row": 1043,
+      "page": 9,
+      "row": 451,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 184320
+          "row": 36864
         }
       },
-      "epoch": 35,
-      "offset": 2048,
-      "documents": 7374503
+      "epoch": 36,
+      "offset": 0,
+      "documents": 7431859
     },
     "openmath": {
-      "page": 1971,
-      "row": 3641,
+      "page": 1987,
+      "row": 1206,
       "start": {
         "file": 18,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 213,
-          "row": 516
+          "row_group": 279,
+          "row": 52
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 8076857
+      "documents": 8139958
     },
     "ultra_qa": {
-      "page": 852,
-      "row": 2758,
+      "page": 859,
+      "row": 1447,
       "start": {
         "file": 6,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 5,
-          "row": 49415
+          "row_group": 6,
+          "row": 13393
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3492550
+      "documents": 3519911
     },
     "ultra_style": {
-      "page": 1301,
-      "row": 1747,
+      "page": 1311,
+      "row": 1812,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 6,
-          "row": 23328
+          "row": 64288
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 5330643
+      "offset": 2048,
+      "documents": 5371668
     },
     "cortex": {
-      "page": 167566,
-      "row": 31,
+      "page": 168873,
+      "row": 98,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 21448479
+      "documents": 21615842
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 640000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "ebc22606ea875e024664b2b5ba82033f919f8d1a96f22dfebda33a7cb5296c83", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "2e88f84323ca4bdc4a8a0fa1302e353f3a442dcb7967001b5b87a164915124f0", "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb", "performance.json": "539995c1e4203016bfc6259363445c7c25dfb51604d74cd18d2cc7ad51bfb71e", "shard_scheduler.json": "bf475a77a1c1a0c722213e8028f3d9758aacd623f550f0da5d55fe94219c3d81", "coverage.json": "33bc21ae73a3caaa60d3061ffb19116a58b5eaf2262d82489a7714a0458299ba", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a", "folder": "step-000000640000"}
+{"step": 645000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "63dedfe6f429aac5ad00d42078291bc517b71b9421a80358aeab37e9ea3770b0", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "09595760b7624b74f118f3a6bb0623c52c6fe05ffe26da58b1c6e977fdb298ab", "training_config.json": "6c3c3ccff89dabb5098a22bc207c2f3abaee8b3bffec3f6cf6ad361a63b1a7fb", "performance.json": "539995c1e4203016bfc6259363445c7c25dfb51604d74cd18d2cc7ad51bfb71e", "shard_scheduler.json": "e3524f5d2e0a02ee5df1bd332f50fbbc4165917e20769e839cc84064c0934bfc", "coverage.json": "63498e908791e3993c2423818a474742357469ed5acc8033f03f8f9c55147e00", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "33a459803c8cffa8929a90def998b4ad2c7c52fd1245365a9ee946b5552d845a", "folder": "step-000000645000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ebc22606ea875e024664b2b5ba82033f919f8d1a96f22dfebda33a7cb5296c83
+oid sha256:63dedfe6f429aac5ad00d42078291bc517b71b9421a80358aeab37e9ea3770b0
 size 360622744

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": 77349.77256644783,
-      "required_pages_per_second_upper_estimate": 0.00535123674234621
+      "required_token_positions_per_second": 77060.14448798755,
+      "required_pages_per_second_upper_estimate": 0.005331199599848598
     },
     {
       "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": 9668.72157080598,
-      "required_pages_per_second_upper_estimate": 0.004048091815830584
+      "required_token_positions_per_second": 9632.518060998444,
+      "required_pages_per_second_upper_estimate": 0.0040329341622894225
     },
     {
       "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": 19337.44314161196,
-      "required_pages_per_second_upper_estimate": 0.004027479443728623
+      "required_token_positions_per_second": 19265.036121996887,
+      "required_pages_per_second_upper_estimate": 0.004012398971044319
     },
     {
       "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": 29006.16471241794,
-      "required_pages_per_second_upper_estimate": 0.001738875284660405
+      "required_token_positions_per_second": 28897.55418299533,
+      "required_pages_per_second_upper_estimate": 0.0017323642492601957
     },
     {
       "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": 29006.16471241794,
-      "required_pages_per_second_upper_estimate": 0.0022551017217554846
+      "required_token_positions_per_second": 28897.55418299533,
+      "required_pages_per_second_upper_estimate": 0.002246657730820108
     },
     {
       "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": 29006.16471241794,
-      "required_pages_per_second_upper_estimate": 0.3720599365377296
+      "required_token_positions_per_second": 28897.55418299533,
+      "required_pages_per_second_upper_estimate": 0.3706667972831971
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.10940042345293932,
-    "producer_p95_seconds": 0.3257678099980694,
-    "gpu_update_seconds": 0.9531766875806063,
-    "producer_batches_per_second": 9.140732443601273,
-    "gpu_batches_per_second": 1.049123434332246,
-    "producer_headroom": 8.712733072652442,
+    "producer_batch_seconds": 0.13888229553117526,
+    "producer_p95_seconds": 0.363497867998376,
+    "gpu_update_seconds": 0.9567591715519431,
+    "producer_batches_per_second": 7.200341815890619,
+    "gpu_batches_per_second": 1.0451951021048658,
+    "producer_headroom": 6.888993070662322,
     "sustainable": true,
-    "buffer_seconds": 60.05013131757819,
+    "buffer_seconds": 60.275827807772416,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 81,
+    "upload_pages_per_commit": 78,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 425795.8908965641,
-    "upload_bytes_per_second": 5026975.921434505,
+    "generation_bytes_per_second": 360738.03359167324,
+    "upload_bytes_per_second": 4634033.565828152,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 91.36588616535848,
+    "estimated_full_shards_per_hour": 77.40598445713661,
     "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": 1.1706882670005143,
-      "bytes": 13271387,
-      "pages": 35,
-      "shards": 4
-    },
-    {
-      "seconds": 0.8168831449984282,
-      "bytes": 1231448,
-      "pages": 36,
-      "shards": 1
+      "seconds": 1.4528134669999417,
+      "bytes": 6484124,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1650056679973204,
-      "bytes": 7481151,
-      "pages": 30,
+      "seconds": 1.400596306000807,
+      "bytes": 7658131,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9575125699993805,
-      "bytes": 1008516,
-      "pages": 30,
+      "seconds": 1.01357181700223,
+      "bytes": 644212,
+      "pages": 18,
       "shards": 1
     },
     {
-      "seconds": 1.5694887619974907,
-      "bytes": 10308114,
-      "pages": 44,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0191816589976952,
-      "bytes": 7135611,
-      "pages": 30,
-      "shards": 3
+      "seconds": 1.5190746830012358,
+      "bytes": 16798507,
+      "pages": 50,
+      "shards": 4
     },
     {
-      "seconds": 1.1397755830003007,
-      "bytes": 6913587,
-      "pages": 33,
+      "seconds": 1.2172141400005785,
+      "bytes": 2428421,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.4925930749996041,
-      "bytes": 12021057,
-      "pages": 40,
-      "shards": 4
-    },
-    {
-      "seconds": 1.3497725520028325,
-      "bytes": 6087212,
+      "seconds": 1.3105249759973958,
+      "bytes": 6190634,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.784925423002278,
-      "bytes": 818137,
-      "pages": 26,
-      "shards": 1
-    },
-    {
-      "seconds": 1.161659340999904,
-      "bytes": 8019004,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0203383750013018,
-      "bytes": 5011407,
-      "pages": 29,
+      "seconds": 1.3242703979994985,
+      "bytes": 5961993,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1040821489987138,
-      "bytes": 11902249,
-      "pages": 35,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1162976039995556,
-      "bytes": 7875044,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.7475975829947856,
+      "bytes": 6736460,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.1005420340006822,
-      "bytes": 1290047,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.445422744996904,
+      "bytes": 6026824,
+      "pages": 21,
+      "shards": 2
     },
     {
-      "seconds": 1.504874573998677,
-      "bytes": 12022604,
-      "pages": 34,
+      "seconds": 1.5578130739959306,
+      "bytes": 13753578,
+      "pages": 49,
       "shards": 4
     },
     {
-      "seconds": 1.3143507820022933,
-      "bytes": 918492,
-      "pages": 27,
+      "seconds": 1.2824578229992767,
+      "bytes": 5780729,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.5125747689999116,
-      "bytes": 6481526,
+      "seconds": 1.419093739001255,
+      "bytes": 6340288,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.7741823629985447,
-      "bytes": 13400964,
-      "pages": 38,
-      "shards": 4
-    },
-    {
-      "seconds": 1.2863249110014294,
-      "bytes": 4967385,
-      "pages": 31,
+      "seconds": 1.3758785059981165,
+      "bytes": 2329519,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.6688778630013985,
-      "bytes": 7600081,
-      "pages": 39,
-      "shards": 3
-    },
-    {
-      "seconds": 1.2229772380014765,
-      "bytes": 1028666,
-      "pages": 30,
-      "shards": 1
+      "seconds": 1.3759045859987964,
+      "bytes": 6265583,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.4301322219980648,
-      "bytes": 6037828,
-      "pages": 34,
+      "seconds": 1.4603665260001435,
+      "bytes": 2072280,
+      "pages": 20,
       "shards": 2
     },
     {
-      "seconds": 1.568642534999526,
-      "bytes": 11958930,
-      "pages": 32,
+      "seconds": 1.5149503700013156,
+      "bytes": 16156661,
+      "pages": 43,
       "shards": 4
     },
     {
-      "seconds": 1.4158421129977796,
-      "bytes": 6969172,
-      "pages": 34,
+      "seconds": 1.2590785899956245,
+      "bytes": 6075199,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.4889061510002648,
-      "bytes": 5773625,
-      "pages": 28,
+      "seconds": 1.3119273870033794,
+      "bytes": 2422074,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.7626394689978042,
-      "bytes": 7617654,
-      "pages": 40,
-      "shards": 3
-    },
-    {
-      "seconds": 1.338195986001665,
-      "bytes": 5044470,
-      "pages": 31,
+      "seconds": 1.4814246340029058,
+      "bytes": 5980643,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.4609920459988643,
-      "bytes": 7624505,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.4380262719932944,
+      "bytes": 12216092,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.1959701189989573,
-      "bytes": 1136148,
-      "pages": 35,
+      "seconds": 1.1287659260051441,
+      "bytes": 715029,
+      "pages": 21,
       "shards": 1
     },
     {
-      "seconds": 1.5574687859989353,
-      "bytes": 11997417,
-      "pages": 38,
+      "seconds": 1.51385825300531,
+      "bytes": 12055505,
+      "pages": 47,
       "shards": 3
     },
     {
-      "seconds": 1.3358469599988894,
-      "bytes": 7186422,
-      "pages": 30,
+      "seconds": 1.379372390001663,
+      "bytes": 7458410,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.527223689998209,
-      "bytes": 6243254,
-      "pages": 33,
+      "seconds": 1.3072389689987176,
+      "bytes": 4796842,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 1.4163199819995498,
-      "bytes": 7360891,
-      "pages": 36,
+      "seconds": 1.831955179994111,
+      "bytes": 7695941,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.1906762940016051,
-      "bytes": 1163380,
-      "pages": 35,
+      "seconds": 1.235046139001497,
+      "bytes": 923574,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.3281080369997653,
-      "bytes": 9814293,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.519131245004246,
+      "bytes": 5792384,
+      "pages": 25,
+      "shards": 2
     },
     {
-      "seconds": 1.3582085800007917,
-      "bytes": 8243145,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.7155198749969713,
+      "bytes": 12980090,
+      "pages": 39,
+      "shards": 4
     },
     {
-      "seconds": 1.29941092199806,
-      "bytes": 6125022,
-      "pages": 36,
+      "seconds": 1.392686164996121,
+      "bytes": 6059365,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.4454559629994037,
-      "bytes": 6189473,
-      "pages": 37,
+      "seconds": 1.2079227729991544,
+      "bytes": 6092047,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.604715366000164,
-      "bytes": 6773206,
+      "seconds": 1.186848571000155,
+      "bytes": 2348648,
       "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.589205835996836,
+      "bytes": 10428035,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.4528134669999417,
-      "bytes": 6484124,
-      "pages": 37,
+      "seconds": 2.1054938710003626,
+      "bytes": 5594661,
+      "pages": 25,
       "shards": 2
     },
     {
-      "seconds": 1.400596306000807,
-      "bytes": 7658131,
+      "seconds": 1.3775813770043897,
+      "bytes": 8401763,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.01357181700223,
-      "bytes": 644212,
-      "pages": 18,
-      "shards": 1
+      "seconds": 1.2358832599929883,
+      "bytes": 6358958,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.5190746830012358,
-      "bytes": 16798507,
-      "pages": 50,
+      "seconds": 1.4860661100028665,
+      "bytes": 11954042,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 1.2172141400005785,
-      "bytes": 2428421,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.2047586019980372,
+      "bytes": 899300,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.3105249759973958,
-      "bytes": 6190634,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.4286277740029618,
+      "bytes": 6408361,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.3242703979994985,
-      "bytes": 5961993,
-      "pages": 28,
+      "seconds": 1.3097595139988698,
+      "bytes": 2126792,
+      "pages": 22,
       "shards": 2
     },
     {
-      "seconds": 1.7475975829947856,
-      "bytes": 6736460,
-      "pages": 41,
+      "seconds": 1.6926589889990282,
+      "bytes": 15912759,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 1.445422744996904,
-      "bytes": 6026824,
-      "pages": 21,
-      "shards": 2
+      "seconds": 1.3871258120052516,
+      "bytes": 7847644,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 1.5578130739959306,
-      "bytes": 13753578,
-      "pages": 49,
-      "shards": 4
+      "seconds": 1.130714740997064,
+      "bytes": 965907,
+      "pages": 28,
+      "shards": 1
     },
     {
-      "seconds": 1.2824578229992767,
-      "bytes": 5780729,
-      "pages": 32,
+      "seconds": 1.9462764029958635,
+      "bytes": 6084966,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.419093739001255,
-      "bytes": 6340288,
+      "seconds": 1.6271315569974831,
+      "bytes": 7351886,
       "pages": 37,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.3758785059981165,
-      "bytes": 2329519,
+      "seconds": 2.2978108899988,
+      "bytes": 11739356,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.3846291969966842,
+      "bytes": 7690785,
+      "pages": 38,
+      "shards": 3
+    },
+    {
+      "seconds": 1.3696625419979682,
+      "bytes": 1334642,
+      "pages": 36,
+      "shards": 1
+    },
+    {
+      "seconds": 1.573794985000859,
+      "bytes": 9980526,
       "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.5847847019977053,
+      "bytes": 2422979,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.3759045859987964,
-      "bytes": 6265583,
+      "seconds": 1.2408236240007682,
+      "bytes": 6154943,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.4603665260001435,
-      "bytes": 2072280,
-      "pages": 20,
-      "shards": 2
+      "seconds": 1.8716023119995953,
+      "bytes": 11724100,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.5149503700013156,
-      "bytes": 16156661,
-      "pages": 43,
-      "shards": 4
+      "seconds": 1.725813149001624,
+      "bytes": 7703275,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.2590785899956245,
-      "bytes": 6075199,
-      "pages": 38,
+      "seconds": 1.292055429003085,
+      "bytes": 5925879,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 1.3119273870033794,
-      "bytes": 2422074,
-      "pages": 29,
-      "shards": 2
+      "seconds": 2.3598094420012785,
+      "bytes": 7376391,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.4814246340029058,
-      "bytes": 5980643,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.3066216919978615,
+      "bytes": 1214010,
+      "pages": 33,
+      "shards": 1
     },
     {
-      "seconds": 1.4380262719932944,
-      "bytes": 12216092,
-      "pages": 40,
+      "seconds": 1.9328020320026553,
+      "bytes": 9830750,
+      "pages": 32,
+      "shards": 3
+    },
+    {
+      "seconds": 1.8036188620026223,
+      "bytes": 13332219,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 1.1287659260051441,
-      "bytes": 715029,
-      "pages": 21,
+      "seconds": 1.2068677400020533,
+      "bytes": 1198185,
+      "pages": 36,
       "shards": 1
     },
     {
-      "seconds": 1.51385825300531,
-      "bytes": 12055505,
-      "pages": 47,
+      "seconds": 1.6970843710005283,
+      "bytes": 7637252,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.379372390001663,
-      "bytes": 7458410,
-      "pages": 31,
+      "seconds": 1.2413891210017027,
+      "bytes": 5260953,
+      "pages": 35,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3634449710007175,
+      "bytes": 5626486,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.4694542010038276,
+      "bytes": 7488941,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.3072389689987176,
-      "bytes": 4796842,
-      "pages": 34,
+      "seconds": 1.3952881760051241,
+      "bytes": 6629850,
+      "pages": 26,
       "shards": 2
+    },
+    {
+      "seconds": 1.629955725002219,
+      "bytes": 12509791,
+      "pages": 40,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2e88f84323ca4bdc4a8a0fa1302e353f3a442dcb7967001b5b87a164915124f0
-size 721776191
+oid sha256:09595760b7624b74f118f3a6bb0623c52c6fe05ffe26da58b1c6e977fdb298ab
+size 721778367