Qwen90M step 960000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 955000,
+  "step": 960000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "5efa97481adb165836eab13ec2e1d8440d32bc3f84314194d651cebe4c6e7d73",
+    "model.safetensors": "4e65901be54350e418ff3e410c921a91fed8b1198c19e5c2d770e8af540c1d48",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "af9b8f970077d4bd0f492458ce4f2991fcef79425d6050931c1082434f1da813",
+    "trainer.pt": "f9543f4cdf1ded5330965f1f4b53c72f52741fe38080a16905aaefc5e88d9075",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4",
-    "shard_scheduler.json": "235d1326e974dcf3b57d55e9bb793f73668d9a67f58d4b3bce1cf5d4eec5e79a",
-    "coverage.json": "471d0cc109ff51b50b59b2ac675520cec412cadab32f02434f02213ec1811ae9",
+    "shard_scheduler.json": "3f9b4cd018de5f1cedb3e083ebbe5af6c595c4b139f36f9ea7fbb7dfa8b447e2",
+    "coverage.json": "5e47ee46906b36cce2cb7784d2cee168630b17f2c4745dc3b4a9441fa676b452",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 955000,
+  "step": 960000,
   "stats": {
-    "climbmix/blocks": 34380000,
-    "climbmix/targets": 57473020907,
-    "climbmix/input_tokens": 57476662882,
-    "ultra_qa/blocks": 12892500,
-    "ultra_qa/targets": 21033125499,
-    "ultra_qa/input_tokens": 21033165912,
-    "cortex/blocks": 12892500,
-    "cortex/targets": 490243480,
-    "cortex/input_tokens": 20493820360,
-    "ultra_style/blocks": 12892500,
-    "ultra_style/targets": 20391585022,
-    "ultra_style/input_tokens": 20392653584,
-    "openmath/blocks": 8595000,
-    "openmath/targets": 13280438962,
-    "openmath/input_tokens": 13284864104,
-    "rewrite6/blocks": 4297500,
-    "rewrite6/targets": 7045113171,
-    "rewrite6/input_tokens": 7052837408
+    "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
   },
   "sources": {
     "climbmix": {
-      "page": 4708,
-      "row": 2416,
+      "page": 4733,
+      "row": 557,
       "start": {
-        "file": 227,
+        "file": 229,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 75,
+          "row_group": 10,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 19286384
+      "offset": 2048,
+      "documents": 19386925
     },
     "rewrite6": {
-      "page": 28,
-      "row": 3208,
+      "page": 42,
+      "row": 3251,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 114688
+          "row": 172032
         }
       },
       "epoch": 53,
-      "offset": 2048,
-      "documents": 11004308
+      "offset": 0,
+      "documents": 11061695
     },
     "openmath": {
-      "page": 2942,
-      "row": 3402,
+      "page": 2958,
+      "row": 998,
       "start": {
         "file": 27,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 260,
-          "row": 886
+          "row_group": 326,
+          "row": 422
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 12053834
+      "offset": 0,
+      "documents": 12116966
     },
     "ultra_qa": {
-      "page": 1272,
-      "row": 1325,
+      "page": 1278,
+      "row": 4021,
       "start": {
         "file": 10,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 0,
-          "row": 13836
+          "row": 38412
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 5211437
+      "documents": 5238709
     },
     "ultra_style": {
-      "page": 1941,
-      "row": 2726,
+      "page": 1951,
+      "row": 2823,
       "start": {
         "file": 11,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 4,
-          "row": 76457
+          "row_group": 5,
+          "row": 32917
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 7953062
+      "offset": 0,
+      "documents": 7994119
     },
     "cortex": {
-      "page": 250033,
-      "row": 97,
+      "page": 251342,
+      "row": 104,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 32004321
+      "documents": 32171880
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 955000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "5efa97481adb165836eab13ec2e1d8440d32bc3f84314194d651cebe4c6e7d73", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "af9b8f970077d4bd0f492458ce4f2991fcef79425d6050931c1082434f1da813", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4", "shard_scheduler.json": "235d1326e974dcf3b57d55e9bb793f73668d9a67f58d4b3bce1cf5d4eec5e79a", "coverage.json": "471d0cc109ff51b50b59b2ac675520cec412cadab32f02434f02213ec1811ae9", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000955000"}
+{"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"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5efa97481adb165836eab13ec2e1d8440d32bc3f84314194d651cebe4c6e7d73
+oid sha256:4e65901be54350e418ff3e410c921a91fed8b1198c19e5c2d770e8af540c1d48
 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": 76445.90440437887,
-      "required_pages_per_second_upper_estimate": 0.005288705045631197
+      "required_token_positions_per_second": 76853.63444472375,
+      "required_pages_per_second_upper_estimate": 0.005316912755886291
     },
     {
       "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": 9555.73805054736,
-      "required_pages_per_second_upper_estimate": 0.0040007879752624945
+      "required_token_positions_per_second": 9606.704305590469,
+      "required_pages_per_second_upper_estimate": 0.004022126481952614
     },
     {
       "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": 19111.47610109472,
-      "required_pages_per_second_upper_estimate": 0.003980416468340475
+      "required_token_positions_per_second": 19213.408611180937,
+      "required_pages_per_second_upper_estimate": 0.004001646322050374
     },
     {
       "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": 28667.214151642078,
-      "required_pages_per_second_upper_estimate": 0.0017185557160894315
+      "required_token_positions_per_second": 28820.112916771406,
+      "required_pages_per_second_upper_estimate": 0.0017277217635960384
     },
     {
       "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": 28667.214151642078,
-      "required_pages_per_second_upper_estimate": 0.002228749806541117
+      "required_token_positions_per_second": 28820.112916771406,
+      "required_pages_per_second_upper_estimate": 0.0022406370129993292
     },
     {
       "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": 28667.214151642078,
-      "required_pages_per_second_upper_estimate": 0.36771224268085423
+      "required_token_positions_per_second": 28820.112916771406,
+      "required_pages_per_second_upper_estimate": 0.36967346387003
     }
   ],
   "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": 0.10924110078121885,
-    "producer_p95_seconds": 0.32970197800023016,
-    "gpu_update_seconds": 0.9644466969740868,
-    "producer_batches_per_second": 9.154063743853484,
-    "gpu_batches_per_second": 1.0368639377764062,
-    "producer_headroom": 8.828606541649735,
+    "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,
     "sustainable": true,
-    "buffer_seconds": 60.76014190936747,
+    "buffer_seconds": 60.43779235113174,
     "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": 340180.5052731563,
-    "upload_bytes_per_second": 5134244.0499124015,
+    "generation_bytes_per_second": 450137.808768403,
+    "upload_bytes_per_second": 4746473.526917738,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 72.99481743474976,
+    "estimated_full_shards_per_hour": 96.58909508980815,
     "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": 1.1959222149998823,
-      "bytes": 7412774,
-      "pages": 30,
-      "shards": 3
-    },
-    {
-      "seconds": 1.488441301000421,
-      "bytes": 6011041,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.0648157450013969,
-      "bytes": 668578,
-      "pages": 20,
-      "shards": 2
-    },
-    {
-      "seconds": 1.9570103289988765,
-      "bytes": 18490744,
-      "pages": 47,
-      "shards": 5
-    },
-    {
-      "seconds": 0.8552882639996824,
-      "bytes": 1175820,
+      "seconds": 1.0184783190015878,
+      "bytes": 1341723,
       "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.1064946840015182,
-      "bytes": 7293995,
-      "pages": 29,
+      "seconds": 1.3343794770007662,
+      "bytes": 7400925,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.736566771996877,
-      "bytes": 6191401,
-      "pages": 38,
+      "seconds": 1.8426530299984734,
+      "bytes": 5834868,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.1331273229989165,
-      "bytes": 4688984,
-      "pages": 31,
+      "seconds": 1.0089386899999226,
+      "bytes": 4885086,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.3685046969985706,
-      "bytes": 7267431,
-      "pages": 27,
-      "shards": 3
+      "seconds": 1.4905629579989181,
+      "bytes": 13725537,
+      "pages": 44,
+      "shards": 4
     },
     {
-      "seconds": 1.4039977110005566,
-      "bytes": 12182760,
-      "pages": 48,
-      "shards": 3
+      "seconds": 1.3237662440005806,
+      "bytes": 894969,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.015954297999997,
-      "bytes": 2619113,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.5763037539982179,
+      "bytes": 10877110,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.3503473790005955,
-      "bytes": 10477535,
-      "pages": 34,
+      "seconds": 1.3618358630010334,
+      "bytes": 7894203,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.9818654609989608,
-      "bytes": 963058,
-      "pages": 27,
+      "seconds": 1.2813600879999285,
+      "bytes": 1187408,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.1789819430014177,
-      "bytes": 7759537,
-      "pages": 39,
+      "seconds": 1.2242876330019499,
+      "bytes": 7412020,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.2661478619993431,
-      "bytes": 11756535,
-      "pages": 35,
+      "seconds": 1.1913529729972652,
+      "bytes": 5109029,
+      "pages": 25,
+      "shards": 2
+    },
+    {
+      "seconds": 1.8349382990018057,
+      "bytes": 11979278,
+      "pages": 43,
       "shards": 3
     },
     {
-      "seconds": 1.2254978609998943,
-      "bytes": 7186634,
+      "seconds": 1.191631180998229,
+      "bytes": 7251240,
       "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.2788624959976005,
-      "bytes": 6210071,
+      "seconds": 1.1317712890013354,
+      "bytes": 6287146,
       "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.412427707000461,
-      "bytes": 887856,
-      "pages": 26,
+      "seconds": 0.9238542080020125,
+      "bytes": 660547,
+      "pages": 18,
       "shards": 1
     },
     {
-      "seconds": 1.2455697520017566,
-      "bytes": 7910132,
-      "pages": 40,
+      "seconds": 1.3592490819974046,
+      "bytes": 11983959,
+      "pages": 49,
+      "shards": 4
+    },
+    {
+      "seconds": 1.0659817249979824,
+      "bytes": 6806858,
+      "pages": 32,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3951733760004572,
+      "bytes": 7437431,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.3195631069975207,
-      "bytes": 10098822,
-      "pages": 36,
+      "seconds": 1.5938844269985566,
+      "bytes": 10568768,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.6743710149967228,
-      "bytes": 8331470,
+      "seconds": 2.7108459590017446,
+      "bytes": 1157065,
       "pages": 33,
-      "shards": 3
+      "shards": 1
     },
     {
-      "seconds": 1.060381152001355,
-      "bytes": 6181629,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.3873778990018764,
+      "bytes": 7922754,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.3323454749988741,
-      "bytes": 6231984,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.5672195150000334,
+      "bytes": 991660,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.3298568760001217,
-      "bytes": 7493698,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.7213281089971133,
+      "bytes": 11058082,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 1.2196332230014377,
-      "bytes": 5954986,
-      "pages": 28,
-      "shards": 2
+      "seconds": 1.1525909759984643,
+      "bytes": 12461215,
+      "pages": 46,
+      "shards": 4
     },
     {
-      "seconds": 1.0563156829994114,
-      "bytes": 932363,
+      "seconds": 1.7147560890007298,
+      "bytes": 951064,
       "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.3606180970018613,
-      "bytes": 7382890,
+      "seconds": 2.651306643998396,
+      "bytes": 1793185,
+      "pages": 11,
+      "shards": 2
+    },
+    {
+      "seconds": 1.1449169079969579,
+      "bytes": 5670243,
       "pages": 28,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.5418479689978994,
-      "bytes": 10959545,
-      "pages": 43,
-      "shards": 3
+      "seconds": 1.2746175409993157,
+      "bytes": 5571788,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 1.2728104929992696,
-      "bytes": 7484326,
-      "pages": 33,
+      "seconds": 1.6674628689979727,
+      "bytes": 7856605,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.4130519919999642,
-      "bytes": 6156286,
+      "seconds": 1.3078784290009935,
+      "bytes": 11982756,
       "pages": 35,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 0.9332762750018446,
-      "bytes": 1158586,
+      "seconds": 0.9895834060007473,
+      "bytes": 1128299,
       "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.577455007001845,
-      "bytes": 11622766,
-      "pages": 35,
+      "seconds": 1.9414047560021572,
+      "bytes": 11291288,
+      "pages": 33,
       "shards": 4
     },
     {
-      "seconds": 1.0520612039981643,
-      "bytes": 811209,
-      "pages": 24,
-      "shards": 1
+      "seconds": 1.799824219000584,
+      "bytes": 6035033,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.4460338999997475,
-      "bytes": 11954296,
-      "pages": 39,
-      "shards": 4
+      "seconds": 1.1105879839997215,
+      "bytes": 2480881,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.3092962549999356,
-      "bytes": 7432740,
-      "pages": 31,
+      "seconds": 1.2019559269974707,
+      "bytes": 10756417,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.1428929149988107,
-      "bytes": 5656732,
-      "pages": 36,
+      "seconds": 1.428911708000669,
+      "bytes": 11976856,
+      "pages": 35,
+      "shards": 3
+    },
+    {
+      "seconds": 1.1643081420006638,
+      "bytes": 2305467,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.3837304669978039,
-      "bytes": 7550021,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.1377382460013905,
+      "bytes": 6053085,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.4462293939977826,
-      "bytes": 810671,
-      "pages": 26,
+      "seconds": 0.9952570979985467,
+      "bytes": 952538,
+      "pages": 29,
       "shards": 1
     },
     {
-      "seconds": 1.2543598109987215,
-      "bytes": 12226155,
-      "pages": 42,
+      "seconds": 1.5738787880000018,
+      "bytes": 11910990,
+      "pages": 41,
+      "shards": 4
+    },
+    {
+      "seconds": 1.6920514609992097,
+      "bytes": 6058111,
+      "pages": 29,
+      "shards": 2
+    },
+    {
+      "seconds": 1.675573243999679,
+      "bytes": 8344893,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.268981666999025,
-      "bytes": 11497273,
-      "pages": 33,
-      "shards": 4
+      "seconds": 1.776490742999158,
+      "bytes": 11106825,
+      "pages": 37,
+      "shards": 3
     },
     {
-      "seconds": 1.0184783190015878,
-      "bytes": 1341723,
-      "pages": 35,
+      "seconds": 1.2736427769996226,
+      "bytes": 1040054,
+      "pages": 31,
       "shards": 1
     },
     {
-      "seconds": 1.3343794770007662,
-      "bytes": 7400925,
-      "pages": 31,
+      "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": 1.3873778990018764,
-      "bytes": 7922754,
-      "pages": 39,
-      "shards": 3
+      "seconds": 1.386075828995672,
+      "bytes": 6310823,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.5672195150000334,
-      "bytes": 991660,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.2331648629988194,
+      "bytes": 6006485,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.7213281089971133,
-      "bytes": 11058082,
-      "pages": 32,
-      "shards": 4
+      "seconds": 1.1499276460017427,
+      "bytes": 7228890,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.1525909759984643,
-      "bytes": 12461215,
-      "pages": 46,
-      "shards": 4
+      "seconds": 1.1513834180004778,
+      "bytes": 6077960,
+      "pages": 38,
+      "shards": 2
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:af9b8f970077d4bd0f492458ce4f2991fcef79425d6050931c1082434f1da813
-size 721766911
+oid sha256:f9543f4cdf1ded5330965f1f4b53c72f52741fe38080a16905aaefc5e88d9075
+size 721772799