Qwen90M step 965000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 960000,
+  "step": 965000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "4e65901be54350e418ff3e410c921a91fed8b1198c19e5c2d770e8af540c1d48",
+    "model.safetensors": "e2aafe51381da07b3941f6b6b93437d13b46fc1b49211291cb65fcfde96084b4",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "f9543f4cdf1ded5330965f1f4b53c72f52741fe38080a16905aaefc5e88d9075",
+    "trainer.pt": "9777eb7df3530cfef81ac2e811a666dd12bace8a24e32e3443b4f91024d5bc91",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4",
-    "shard_scheduler.json": "3f9b4cd018de5f1cedb3e083ebbe5af6c595c4b139f36f9ea7fbb7dfa8b447e2",
-    "coverage.json": "5e47ee46906b36cce2cb7784d2cee168630b17f2c4745dc3b4a9441fa676b452",
+    "shard_scheduler.json": "70835f6e96b0eba26f20e8b0362c9f569b098fd421beee58fe9217b58e98c31a",
+    "coverage.json": "844a83043724bea786727949d8ab88d794ec4ad1927dcb688a87fd013a2e7458",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 960000,
+  "step": 965000,
   "stats": {
-    "climbmix/blocks": 34560000,
-    "climbmix/targets": 57774024820,
-    "climbmix/input_tokens": 57777685720,
-    "ultra_qa/blocks": 12960000,
-    "ultra_qa/targets": 21143110387,
-    "ultra_qa/input_tokens": 21143151007,
-    "cortex/blocks": 12960000,
-    "cortex/targets": 492808988,
-    "cortex/input_tokens": 20601124207,
-    "ultra_style/blocks": 12960000,
-    "ultra_style/targets": 20498872762,
-    "ultra_style/input_tokens": 20499946726,
-    "openmath/blocks": 8640000,
-    "openmath/targets": 13350009288,
-    "openmath/input_tokens": 13354457630,
-    "rewrite6/blocks": 4320000,
-    "rewrite6/targets": 7082059193,
-    "rewrite6/input_tokens": 7089823666
+    "climbmix/blocks": 34740000,
+    "climbmix/targets": 58074898975,
+    "climbmix/input_tokens": 58078578876,
+    "ultra_qa/blocks": 13027500,
+    "ultra_qa/targets": 21253284217,
+    "ultra_qa/input_tokens": 21253325041,
+    "cortex/blocks": 13027500,
+    "cortex/targets": 495374770,
+    "cortex/input_tokens": 20708481919,
+    "ultra_style/blocks": 13027500,
+    "ultra_style/targets": 20605371821,
+    "ultra_style/input_tokens": 20606451606,
+    "openmath/blocks": 8685000,
+    "openmath/targets": 13419500477,
+    "openmath/input_tokens": 13423972043,
+    "rewrite6/blocks": 4342500,
+    "rewrite6/targets": 7118963627,
+    "rewrite6/input_tokens": 7126768533
   },
   "sources": {
     "climbmix": {
-      "page": 4733,
-      "row": 557,
+      "page": 4757,
+      "row": 3358,
       "start": {
-        "file": 229,
+        "file": 230,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 10,
+          "row_group": 23,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 19386925
+      "offset": 12288,
+      "documents": 19488030
     },
     "rewrite6": {
-      "page": 42,
-      "row": 3251,
+      "page": 6,
+      "row": 2806,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 172032
+          "row": 24576
         }
       },
-      "epoch": 53,
+      "epoch": 54,
       "offset": 0,
-      "documents": 11061695
+      "documents": 11119198
     },
     "openmath": {
-      "page": 2958,
-      "row": 998,
+      "page": 2973,
+      "row": 2684,
       "start": {
         "file": 27,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 326,
-          "row": 422
+          "row_group": 387,
+          "row": 862
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 12116966
+      "documents": 12180092
     },
     "ultra_qa": {
-      "page": 1278,
-      "row": 4021,
+      "page": 1285,
+      "row": 2671,
       "start": {
         "file": 10,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 0,
-          "row": 38412
+          "row_group": 1,
+          "row": 2584
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 5238709
+      "documents": 5266031
     },
     "ultra_style": {
-      "page": 1951,
-      "row": 2823,
+      "page": 1961,
+      "row": 4055,
       "start": {
         "file": 11,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 5,
-          "row": 32917
+          "row": 73877
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 7994119
+      "offset": 4096,
+      "documents": 8036311
     },
     "cortex": {
-      "page": 251342,
-      "row": 104,
+      "page": 252652,
+      "row": 24,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 32171880
+      "documents": 32339480
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 960000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "4e65901be54350e418ff3e410c921a91fed8b1198c19e5c2d770e8af540c1d48", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "f9543f4cdf1ded5330965f1f4b53c72f52741fe38080a16905aaefc5e88d9075", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4", "shard_scheduler.json": "3f9b4cd018de5f1cedb3e083ebbe5af6c595c4b139f36f9ea7fbb7dfa8b447e2", "coverage.json": "5e47ee46906b36cce2cb7784d2cee168630b17f2c4745dc3b4a9441fa676b452", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000960000"}
+{"step": 965000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e2aafe51381da07b3941f6b6b93437d13b46fc1b49211291cb65fcfde96084b4", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "9777eb7df3530cfef81ac2e811a666dd12bace8a24e32e3443b4f91024d5bc91", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4", "shard_scheduler.json": "70835f6e96b0eba26f20e8b0362c9f569b098fd421beee58fe9217b58e98c31a", "coverage.json": "844a83043724bea786727949d8ab88d794ec4ad1927dcb688a87fd013a2e7458", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000965000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4e65901be54350e418ff3e410c921a91fed8b1198c19e5c2d770e8af540c1d48
+oid sha256:e2aafe51381da07b3941f6b6b93437d13b46fc1b49211291cb65fcfde96084b4
 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": 76853.63444472375,
-      "required_pages_per_second_upper_estimate": 0.005316912755886291
+      "required_token_positions_per_second": 76671.34370399594,
+      "required_pages_per_second_upper_estimate": 0.005304301459469948
     },
     {
       "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": 9606.704305590469,
-      "required_pages_per_second_upper_estimate": 0.004022126481952614
+      "required_token_positions_per_second": 9583.917962999492,
+      "required_pages_per_second_upper_estimate": 0.004012586316142715
     },
     {
       "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": 19213.408611180937,
-      "required_pages_per_second_upper_estimate": 0.004001646322050374
+      "required_token_positions_per_second": 19167.835925998985,
+      "required_pages_per_second_upper_estimate": 0.003992154733559502
     },
     {
       "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": 28820.112916771406,
-      "required_pages_per_second_upper_estimate": 0.0017277217635960384
+      "required_token_positions_per_second": 28751.75388899848,
+      "required_pages_per_second_upper_estimate": 0.001723623744259243
     },
     {
       "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": 28820.112916771406,
-      "required_pages_per_second_upper_estimate": 0.0022406370129993292
+      "required_token_positions_per_second": 28751.75388899848,
+      "required_pages_per_second_upper_estimate": 0.0022353223992695706
     },
     {
       "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": 28820.112916771406,
-      "required_pages_per_second_upper_estimate": 0.36967346387003
+      "required_token_positions_per_second": 28751.75388899848,
+      "required_pages_per_second_upper_estimate": 0.36879662765996435
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,409 +158,409 @@
     "prefetch_batches": 63,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 720750,
-    "producer_batch_seconds": 0.13042824582839785,
-    "producer_p95_seconds": 0.3381571759964572,
-    "gpu_update_seconds": 0.9593300373195515,
-    "producer_batches_per_second": 7.667050903342536,
-    "gpu_batches_per_second": 1.0423941303809103,
-    "producer_headroom": 7.355232229234495,
+    "producer_batch_seconds": 0.09754042230423465,
+    "producer_p95_seconds": 0.34494055299728643,
+    "gpu_update_seconds": 0.961610902303222,
+    "producer_batches_per_second": 10.252159836677123,
+    "gpu_batches_per_second": 1.0399216539712992,
+    "producer_headroom": 9.858588671103941,
     "sustainable": true,
-    "buffer_seconds": 60.43779235113174,
+    "buffer_seconds": 60.58148684510299,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 82,
+    "upload_pages_per_commit": 87,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 450137.808768403,
-    "upload_bytes_per_second": 4746473.526917738,
+    "generation_bytes_per_second": 366223.02480693226,
+    "upload_bytes_per_second": 5080294.423461347,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 96.58909508980815,
+    "estimated_full_shards_per_hour": 78.58293588787056,
     "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": 1.0184783190015878,
-      "bytes": 1341723,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.675573243999679,
+      "bytes": 8344893,
+      "pages": 36,
+      "shards": 3
+    },
+    {
+      "seconds": 1.776490742999158,
+      "bytes": 11106825,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.3343794770007662,
-      "bytes": 7400925,
+      "seconds": 1.2736427769996226,
+      "bytes": 1040054,
       "pages": 31,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3692391859985946,
+      "bytes": 8054475,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.8426530299984734,
-      "bytes": 5834868,
-      "pages": 33,
+      "seconds": 1.189110659997823,
+      "bytes": 6011923,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0089386899999226,
-      "bytes": 4885086,
-      "pages": 33,
+      "seconds": 0.9417887669987977,
+      "bytes": 2628206,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.4905629579989181,
-      "bytes": 13725537,
-      "pages": 44,
+      "seconds": 1.5466363100022136,
+      "bytes": 15617223,
+      "pages": 32,
       "shards": 4
     },
     {
-      "seconds": 1.3237662440005806,
-      "bytes": 894969,
-      "pages": 26,
+      "seconds": 1.0353363290014386,
+      "bytes": 1240846,
+      "pages": 36,
       "shards": 1
     },
     {
-      "seconds": 1.5763037539982179,
-      "bytes": 10877110,
-      "pages": 36,
+      "seconds": 1.4493512070002907,
+      "bytes": 7603435,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.3618358630010334,
-      "bytes": 7894203,
-      "pages": 37,
+      "seconds": 1.3544745270010026,
+      "bytes": 10741434,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.2813600879999285,
-      "bytes": 1187408,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.3606159429982654,
+      "bytes": 2419629,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.2242876330019499,
-      "bytes": 7412020,
-      "pages": 31,
-      "shards": 3
+      "seconds": 1.757866293999541,
+      "bytes": 6331179,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.1913529729972652,
-      "bytes": 5109029,
-      "pages": 25,
-      "shards": 2
+      "seconds": 1.503990763994807,
+      "bytes": 11868685,
+      "pages": 38,
+      "shards": 3
     },
     {
-      "seconds": 1.8349382990018057,
-      "bytes": 11979278,
-      "pages": 43,
+      "seconds": 1.0632484200032195,
+      "bytes": 2451644,
+      "pages": 28,
       "shards": 3
     },
     {
-      "seconds": 1.191631180998229,
-      "bytes": 7251240,
-      "pages": 35,
+      "seconds": 1.337224768998567,
+      "bytes": 9767269,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.1317712890013354,
-      "bytes": 6287146,
+      "seconds": 1.0372320719980053,
+      "bytes": 2714870,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 0.9238542080020125,
-      "bytes": 660547,
-      "pages": 18,
-      "shards": 1
-    },
-    {
-      "seconds": 1.3592490819974046,
-      "bytes": 11983959,
-      "pages": 49,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0659817249979824,
-      "bytes": 6806858,
-      "pages": 32,
+      "seconds": 1.8995327459997497,
+      "bytes": 6052441,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.3951733760004572,
-      "bytes": 7437431,
-      "pages": 31,
+      "seconds": 1.377700143995753,
+      "bytes": 10634260,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.5938844269985566,
-      "bytes": 10568768,
-      "pages": 34,
+      "seconds": 1.4519578809995437,
+      "bytes": 8113266,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 2.7108459590017446,
-      "bytes": 1157065,
-      "pages": 33,
-      "shards": 1
+      "seconds": 1.386075828995672,
+      "bytes": 6310823,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.3873778990018764,
-      "bytes": 7922754,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.2331648629988194,
+      "bytes": 6006485,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.5672195150000334,
-      "bytes": 991660,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.1499276460017427,
+      "bytes": 7228890,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.7213281089971133,
-      "bytes": 11058082,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.1513834180004778,
+      "bytes": 6077960,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.1525909759984643,
-      "bytes": 12461215,
-      "pages": 46,
-      "shards": 4
+      "seconds": 1.0119541929961997,
+      "bytes": 2485722,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.7147560890007298,
-      "bytes": 951064,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.1930818900000304,
+      "bytes": 5619417,
+      "pages": 14,
+      "shards": 2
     },
     {
-      "seconds": 2.651306643998396,
-      "bytes": 1793185,
-      "pages": 11,
+      "seconds": 1.3075707610041718,
+      "bytes": 7296914,
+      "pages": 42,
       "shards": 2
     },
     {
-      "seconds": 1.1449169079969579,
-      "bytes": 5670243,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.5529981709987624,
+      "bytes": 11067080,
+      "pages": 38,
+      "shards": 4
     },
     {
-      "seconds": 1.2746175409993157,
-      "bytes": 5571788,
-      "pages": 31,
+      "seconds": 1.1875661560043227,
+      "bytes": 6040644,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.6674628689979727,
-      "bytes": 7856605,
-      "pages": 39,
-      "shards": 3
+      "seconds": 0.9448312890017405,
+      "bytes": 1217674,
+      "pages": 36,
+      "shards": 1
     },
     {
-      "seconds": 1.3078784290009935,
-      "bytes": 11982756,
-      "pages": 35,
-      "shards": 3
+      "seconds": 2.036403302998224,
+      "bytes": 12487493,
+      "pages": 38,
+      "shards": 4
     },
     {
-      "seconds": 0.9895834060007473,
-      "bytes": 1128299,
-      "pages": 35,
+      "seconds": 0.962299234000966,
+      "bytes": 720156,
+      "pages": 23,
       "shards": 1
     },
     {
-      "seconds": 1.9414047560021572,
-      "bytes": 11291288,
-      "pages": 33,
+      "seconds": 1.5615880700061098,
+      "bytes": 13769456,
+      "pages": 47,
       "shards": 4
     },
     {
-      "seconds": 1.799824219000584,
-      "bytes": 6035033,
-      "pages": 37,
+      "seconds": 1.0934254949970637,
+      "bytes": 6083579,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1105879839997215,
-      "bytes": 2480881,
-      "pages": 30,
+      "seconds": 1.342605496996839,
+      "bytes": 1094629,
+      "pages": 34,
+      "shards": 1
+    },
+    {
+      "seconds": 1.425732910996885,
+      "bytes": 11302182,
+      "pages": 41,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1526390029976028,
+      "bytes": 5901438,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2019559269974707,
-      "bytes": 10756417,
-      "pages": 40,
+      "seconds": 1.9387289110018173,
+      "bytes": 8423867,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.428911708000669,
-      "bytes": 11976856,
+      "seconds": 1.6871719589980785,
+      "bytes": 10804637,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.1643081420006638,
-      "bytes": 2305467,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1380872770023416,
+      "bytes": 998627,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.1377382460013905,
-      "bytes": 6053085,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.2768399569977191,
+      "bytes": 7586837,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 0.9952570979985467,
-      "bytes": 952538,
+      "seconds": 1.1069597620007698,
+      "bytes": 5924616,
       "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.2267345689979265,
+      "bytes": 1209498,
+      "pages": 33,
       "shards": 1
     },
     {
-      "seconds": 1.5738787880000018,
-      "bytes": 11910990,
-      "pages": 41,
+      "seconds": 1.4170462420006515,
+      "bytes": 11019556,
+      "pages": 29,
       "shards": 4
     },
     {
-      "seconds": 1.6920514609992097,
-      "bytes": 6058111,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.1776043950012536,
+      "bytes": 12312597,
+      "pages": 48,
+      "shards": 3
     },
     {
-      "seconds": 1.675573243999679,
-      "bytes": 8344893,
-      "pages": 36,
+      "seconds": 1.1231176639994374,
+      "bytes": 2410876,
+      "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.776490742999158,
-      "bytes": 11106825,
-      "pages": 37,
+      "seconds": 1.1969935200031614,
+      "bytes": 10533402,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.2736427769996226,
-      "bytes": 1040054,
-      "pages": 31,
+      "seconds": 0.9813283670009696,
+      "bytes": 889426,
+      "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.3692391859985946,
-      "bytes": 8054475,
-      "pages": 40,
+      "seconds": 1.3786637550001615,
+      "bytes": 7505060,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.189110659997823,
-      "bytes": 6011923,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.9612054499957594,
+      "bytes": 587735,
+      "pages": 19,
+      "shards": 1
     },
     {
-      "seconds": 0.9417887669987977,
-      "bytes": 2628206,
-      "pages": 38,
+      "seconds": 1.4943926799969631,
+      "bytes": 17356969,
+      "pages": 43,
+      "shards": 5
+    },
+    {
+      "seconds": 1.3928040120008518,
+      "bytes": 6228637,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.5466363100022136,
-      "bytes": 15617223,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.2593115519994171,
+      "bytes": 6168643,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.0353363290014386,
-      "bytes": 1240846,
-      "pages": 36,
-      "shards": 1
+      "seconds": 1.411603455999284,
+      "bytes": 2470060,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.4493512070002907,
-      "bytes": 7603435,
+      "seconds": 1.2517306329973508,
+      "bytes": 10612343,
       "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.3544745270010026,
-      "bytes": 10741434,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.098978062000242,
+      "bytes": 637051,
+      "pages": 18,
+      "shards": 1
     },
     {
-      "seconds": 1.3606159429982654,
-      "bytes": 2419629,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.4846390739985509,
+      "bytes": 13365694,
+      "pages": 41,
+      "shards": 4
     },
     {
-      "seconds": 1.757866293999541,
-      "bytes": 6331179,
+      "seconds": 1.3245377540006302,
+      "bytes": 6089742,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.503990763994807,
-      "bytes": 11868685,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0632484200032195,
-      "bytes": 2451644,
-      "pages": 28,
-      "shards": 3
-    },
-    {
-      "seconds": 1.337224768998567,
-      "bytes": 9767269,
+      "seconds": 1.1078838069952326,
+      "bytes": 6470653,
       "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0372320719980053,
-      "bytes": 2714870,
-      "pages": 38,
+      "seconds": 1.8197720390016912,
+      "bytes": 6211915,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.8995327459997497,
-      "bytes": 6052441,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.0536710650048917,
+      "bytes": 923281,
+      "pages": 30,
+      "shards": 1
     },
     {
-      "seconds": 1.377700143995753,
-      "bytes": 10634260,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.7383342659959453,
+      "bytes": 12982783,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 1.4519578809995437,
-      "bytes": 8113266,
-      "pages": 31,
+      "seconds": 1.1493779860029463,
+      "bytes": 10545016,
+      "pages": 43,
       "shards": 3
     },
     {
-      "seconds": 1.386075828995672,
-      "bytes": 6310823,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2331648629988194,
-      "bytes": 6006485,
+      "seconds": 1.2370521369957714,
+      "bytes": 2360341,
       "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.1499276460017427,
-      "bytes": 7228890,
-      "pages": 36,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1513834180004778,
-      "bytes": 6077960,
-      "pages": 38,
+      "seconds": 1.5772486200003186,
+      "bytes": 6280980,
+      "pages": 37,
       "shards": 2
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f9543f4cdf1ded5330965f1f4b53c72f52741fe38080a16905aaefc5e88d9075
-size 721772799
+oid sha256:9777eb7df3530cfef81ac2e811a666dd12bace8a24e32e3443b4f91024d5bc91
+size 721767103