Qwen90M step 755000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 750000,
+  "step": 755000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "967a5dad40471a75696936f82686db770fe83a31472a4295ac8a0745743e376e",
+    "model.safetensors": "75cc8e7df2946f47615bba735e2412ba7d26acf35ff8d177274e2b278117ccf7",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "fa840d52fec48122d7ba48ef7ead58d0b3533e834c3dec925aa67b3f13281da5",
+    "trainer.pt": "04389db27054d0deb18e21ea9848cc49ae3250a62872254d899b5d9fb8463946",
     "training_config.json": "75248af0a321738b405c3d51d1d55c7040dbc0785759d95f8dbdf6cd071a6687",
     "performance.json": "48e3aa7938b07d5e08810d60bc208cc2cb9133c4c39a24151758b79306675379",
-    "shard_scheduler.json": "0b3d702d2793873e91a091708c94a6303900163fa993485826b15635a9c69670",
-    "coverage.json": "456e31956d090f2507fe86424334753af439d32400f9d8e6814c0eb6de160a79",
+    "shard_scheduler.json": "6b4bb13b0a72c739a75dbb27075aa6cb85d8a6601c3ec79525308daa070d0808",
+    "coverage.json": "550b419d22a23b028c27e10458eb1b3ca5486992854436adfd61260ee070e83f",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "13510bfc04d8fed598fd0d09c1f778fbaebf19cefdd81f23f5ef3b909ef94a63"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 750000,
+  "step": 755000,
   "stats": {
-    "climbmix/blocks": 27000000,
-    "climbmix/targets": 45135657589,
-    "climbmix/input_tokens": 45138517546,
-    "ultra_qa/blocks": 10125000,
-    "ultra_qa/targets": 16517745406,
-    "ultra_qa/input_tokens": 16517777245,
-    "cortex/blocks": 10125000,
-    "cortex/targets": 384982808,
-    "cortex/input_tokens": 16094233831,
-    "ultra_style/blocks": 10125000,
-    "ultra_style/targets": 16014758015,
-    "ultra_style/input_tokens": 16015597822,
-    "openmath/blocks": 6750000,
-    "openmath/targets": 10429991701,
-    "openmath/input_tokens": 10433466669,
-    "rewrite6/blocks": 3375000,
-    "rewrite6/targets": 5532858815,
-    "rewrite6/input_tokens": 5538924694
+    "climbmix/blocks": 27180000,
+    "climbmix/targets": 45436583708,
+    "climbmix/input_tokens": 45439462550,
+    "ultra_qa/blocks": 10192500,
+    "ultra_qa/targets": 16627829240,
+    "ultra_qa/input_tokens": 16627861293,
+    "cortex/blocks": 10192500,
+    "cortex/targets": 387543241,
+    "cortex/input_tokens": 16201531922,
+    "ultra_style/blocks": 10192500,
+    "ultra_style/targets": 16121676984,
+    "ultra_style/input_tokens": 16122522199,
+    "openmath/blocks": 6795000,
+    "openmath/targets": 10499419459,
+    "openmath/input_tokens": 10502917378,
+    "rewrite6/blocks": 3397500,
+    "rewrite6/targets": 5569717202,
+    "rewrite6/input_tokens": 5575823779
   },
   "sources": {
     "climbmix": {
-      "page": 3697,
-      "row": 2703,
+      "page": 3722,
+      "row": 698,
       "start": {
-        "file": 178,
+        "file": 180,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 79,
+          "row_group": 14,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 16384,
-      "documents": 15145615
+      "offset": 2048,
+      "documents": 15246010
     },
     "rewrite6": {
-      "page": 3,
-      "row": 2523,
+      "page": 17,
+      "row": 3070,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 12288
+          "row": 69632
         }
       },
       "epoch": 42,
       "offset": 0,
-      "documents": 8641779
+      "documents": 8699670
     },
     "openmath": {
-      "page": 2310,
-      "row": 3999,
+      "page": 2326,
+      "row": 1294,
       "start": {
         "file": 21,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 292,
-          "row": 110
+          "row_group": 357,
+          "row": 646
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 9465759
+      "documents": 9528590
     },
     "ultra_qa": {
-      "page": 999,
-      "row": 1072,
+      "page": 1005,
+      "row": 3733,
       "start": {
         "file": 7,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 7,
-          "row": 2368
+          "row": 26944
         }
       },
       "epoch": 0,
       "offset": 4096,
-      "documents": 4092976
+      "documents": 4120213
     },
     "ultra_style": {
-      "page": 1525,
-      "row": 454,
+      "page": 1535,
+      "row": 679,
       "start": {
         "file": 9,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 0,
-          "row": 81926
+          "row_group": 1,
+          "row": 37522
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 6246854
+      "documents": 6288039
     },
     "cortex": {
-      "page": 196363,
-      "row": 45,
+      "page": 197671,
+      "row": 117,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 25134509
+      "documents": 25302005
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 750000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "967a5dad40471a75696936f82686db770fe83a31472a4295ac8a0745743e376e", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "fa840d52fec48122d7ba48ef7ead58d0b3533e834c3dec925aa67b3f13281da5", "training_config.json": "75248af0a321738b405c3d51d1d55c7040dbc0785759d95f8dbdf6cd071a6687", "performance.json": "48e3aa7938b07d5e08810d60bc208cc2cb9133c4c39a24151758b79306675379", "shard_scheduler.json": "0b3d702d2793873e91a091708c94a6303900163fa993485826b15635a9c69670", "coverage.json": "456e31956d090f2507fe86424334753af439d32400f9d8e6814c0eb6de160a79", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "13510bfc04d8fed598fd0d09c1f778fbaebf19cefdd81f23f5ef3b909ef94a63", "folder": "step-000000750000"}
+{"step": 755000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "75cc8e7df2946f47615bba735e2412ba7d26acf35ff8d177274e2b278117ccf7", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "04389db27054d0deb18e21ea9848cc49ae3250a62872254d899b5d9fb8463946", "training_config.json": "75248af0a321738b405c3d51d1d55c7040dbc0785759d95f8dbdf6cd071a6687", "performance.json": "48e3aa7938b07d5e08810d60bc208cc2cb9133c4c39a24151758b79306675379", "shard_scheduler.json": "6b4bb13b0a72c739a75dbb27075aa6cb85d8a6601c3ec79525308daa070d0808", "coverage.json": "550b419d22a23b028c27e10458eb1b3ca5486992854436adfd61260ee070e83f", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "13510bfc04d8fed598fd0d09c1f778fbaebf19cefdd81f23f5ef3b909ef94a63", "folder": "step-000000755000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:967a5dad40471a75696936f82686db770fe83a31472a4295ac8a0745743e376e
+oid sha256:75cc8e7df2946f47615bba735e2412ba7d26acf35ff8d177274e2b278117ccf7
 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": 76589.24498588534,
-      "required_pages_per_second_upper_estimate": 0.0052986216796557855
+      "required_token_positions_per_second": 76927.36162945612,
+      "required_pages_per_second_upper_estimate": 0.005322013373596714
     },
     {
       "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": 9573.655623235667,
-      "required_pages_per_second_upper_estimate": 0.004008289688785624
+      "required_token_positions_per_second": 9615.920203682015,
+      "required_pages_per_second_upper_estimate": 0.0040259849860337085
     },
     {
       "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": 19147.311246471334,
-      "required_pages_per_second_upper_estimate": 0.00398787998408609
+      "required_token_positions_per_second": 19231.84040736403,
+      "required_pages_per_second_upper_estimate": 0.00400548517911616
     },
     {
       "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": 28720.966869707,
-      "required_pages_per_second_upper_estimate": 0.0017217781094617756
+      "required_token_positions_per_second": 28847.760611046047,
+      "required_pages_per_second_upper_estimate": 0.0017293792006522182
     },
     {
       "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": 28720.966869707,
-      "required_pages_per_second_upper_estimate": 0.0022329288439374446
+      "required_token_positions_per_second": 28847.760611046047,
+      "required_pages_per_second_upper_estimate": 0.002242786499619828
     },
     {
       "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": 28720.966869707,
-      "required_pages_per_second_upper_estimate": 0.3684017248330191
+      "required_token_positions_per_second": 28847.760611046047,
+      "required_pages_per_second_upper_estimate": 0.3700280988064038
     }
   ],
   "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": 0.10013397956231529,
-    "producer_p95_seconds": 0.27769846199953463,
-    "gpu_update_seconds": 0.962641687009545,
-    "producer_batches_per_second": 9.986619970273736,
-    "gpu_batches_per_second": 1.0388081188406757,
-    "producer_headroom": 9.613536695707522,
+    "producer_batch_seconds": 0.11758859633584962,
+    "producer_p95_seconds": 0.2860190520004835,
+    "gpu_update_seconds": 0.9584106153949902,
+    "producer_batches_per_second": 8.504226014773227,
+    "gpu_batches_per_second": 1.0433941193231353,
+    "producer_headroom": 8.150540488276892,
     "sustainable": true,
-    "buffer_seconds": 60.646426281601336,
+    "buffer_seconds": 60.37986876988438,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 83,
+    "upload_pages_per_commit": 86,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 288485.1748269605,
-    "upload_bytes_per_second": 5253353.013598868,
+    "generation_bytes_per_second": 451012.4654282901,
+    "upload_bytes_per_second": 5097647.062415216,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 61.90220292669879,
+    "estimated_full_shards_per_hour": 96.77677604805496,
     "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": 1.3444492599992373,
-      "bytes": 6214787,
-      "pages": 36,
+      "seconds": 1.0973008010005287,
+      "bytes": 11080849,
+      "pages": 33,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1580557140005112,
+      "bytes": 6155625,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.2648084209977242,
-      "bytes": 5986982,
+      "seconds": 0.8461053749997518,
+      "bytes": 1155308,
+      "pages": 32,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2314064579986734,
+      "bytes": 7429916,
       "pages": 31,
-      "shards": 2
+      "shards": 3
     },
     {
-      "seconds": 1.4115151900005003,
-      "bytes": 7731251,
+      "seconds": 1.4280758529966988,
+      "bytes": 11440670,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.1964002810018428,
-      "bytes": 11692696,
-      "pages": 29,
+      "seconds": 1.364633949000563,
+      "bytes": 7290084,
+      "pages": 30,
       "shards": 3
     },
     {
-      "seconds": 1.0258839710004395,
-      "bytes": 2700953,
-      "pages": 38,
+      "seconds": 1.0805251250021684,
+      "bytes": 6054722,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2136837799989735,
-      "bytes": 9835275,
-      "pages": 34,
-      "shards": 3
+      "seconds": 2.882830210000975,
+      "bytes": 1047147,
+      "pages": 31,
+      "shards": 1
     },
     {
-      "seconds": 1.2959440570011793,
-      "bytes": 5942943,
-      "pages": 35,
-      "shards": 2
+      "seconds": 1.6690692930023943,
+      "bytes": 11069831,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 0.9916904240017175,
-      "bytes": 2484673,
-      "pages": 30,
+      "seconds": 2.282538455998292,
+      "bytes": 6276829,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.0907142409996595,
-      "bytes": 6115368,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.4166184950008756,
+      "bytes": 8285024,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.3705160650024482,
-      "bytes": 13033658,
-      "pages": 37,
-      "shards": 4
+      "seconds": 1.271756786001788,
+      "bytes": 10114603,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.2911444290002692,
-      "bytes": 5949225,
+      "seconds": 1.2068859079990943,
+      "bytes": 6031296,
+      "pages": 36,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0317238650022773,
+      "bytes": 2334152,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.0684557280001172,
-      "bytes": 6155327,
-      "pages": 37,
+      "seconds": 1.198212211002101,
+      "bytes": 5956669,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2350120470000547,
-      "bytes": 2680827,
-      "pages": 36,
+      "seconds": 1.182239486999606,
+      "bytes": 5356316,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.5306445749993145,
-      "bytes": 9835087,
+      "seconds": 1.588657971999055,
+      "bytes": 13335657,
       "pages": 33,
+      "shards": 4
+    },
+    {
+      "seconds": 0.8740047409992258,
+      "bytes": 947216,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.4317332190003071,
+      "bytes": 7796978,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9580864189993008,
-      "bytes": 989214,
-      "pages": 29,
+      "seconds": 1.0188206910024746,
+      "bytes": 812811,
+      "pages": 23,
       "shards": 1
     },
     {
-      "seconds": 1.3733238169988908,
-      "bytes": 13560061,
-      "pages": 41,
-      "shards": 4
+      "seconds": 1.1825939109985484,
+      "bytes": 10557366,
+      "pages": 46,
+      "shards": 3
     },
     {
-      "seconds": 1.2763601959995867,
-      "bytes": 5814992,
-      "pages": 27,
+      "seconds": 1.6627466680001817,
+      "bytes": 7279837,
+      "pages": 30,
+      "shards": 3
+    },
+    {
+      "seconds": 1.578546805998485,
+      "bytes": 6985484,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.203305928996997,
-      "bytes": 7121753,
+      "seconds": 1.4114699799974915,
+      "bytes": 10209990,
       "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.1241114900003595,
-      "bytes": 6180535,
-      "pages": 36,
+      "seconds": 1.0779238770010124,
+      "bytes": 1827152,
+      "pages": 11,
       "shards": 2
     },
     {
-      "seconds": 1.1907369090004067,
-      "bytes": 6141214,
-      "pages": 35,
+      "seconds": 0.9918526799992833,
+      "bytes": 6095255,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 0.9464336839992029,
-      "bytes": 2380354,
-      "pages": 28,
+      "seconds": 1.0435619140007475,
+      "bytes": 6434451,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.6319502810001723,
-      "bytes": 16693932,
-      "pages": 40,
-      "shards": 4
+      "seconds": 1.1101431559982302,
+      "bytes": 7100100,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 0.91435248900234,
-      "bytes": 2656824,
+      "seconds": 1.0310441840010753,
+      "bytes": 5933626,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1157520460001251,
-      "bytes": 5879867,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1945516140003747,
-      "bytes": 6236357,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.339776704000542,
+      "bytes": 13443293,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 1.1350728899997193,
-      "bytes": 701024,
-      "pages": 21,
+      "seconds": 1.7319635569983802,
+      "bytes": 934829,
+      "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.6311224940000102,
-      "bytes": 11525749,
-      "pages": 46,
-      "shards": 4
-    },
-    {
-      "seconds": 1.0188665900022897,
-      "bytes": 6807685,
-      "pages": 33,
+      "seconds": 1.0800791440015018,
+      "bytes": 6107947,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1573900719995436,
-      "bytes": 7406360,
-      "pages": 30,
+      "seconds": 1.2829358429989952,
+      "bytes": 6479810,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.3309355630008213,
-      "bytes": 11158362,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.002878891002183,
+      "bytes": 6148473,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 0.9890253090015904,
-      "bytes": 2755782,
+      "seconds": 1.581660068997735,
+      "bytes": 6199462,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3326452839974081,
-      "bytes": 5962377,
+      "seconds": 1.3033671679986583,
+      "bytes": 13020043,
+      "pages": 37,
+      "shards": 4
+    },
+    {
+      "seconds": 1.1211934040002234,
+      "bytes": 5810769,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.016969195999991,
-      "bytes": 1171347,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.2197048900015943,
+      "bytes": 2772365,
+      "pages": 38,
+      "shards": 2
     },
     {
-      "seconds": 1.6574927980000211,
-      "bytes": 17383397,
-      "pages": 38,
-      "shards": 5
+      "seconds": 1.9406323449984484,
+      "bytes": 5788082,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.1213059549991158,
-      "bytes": 6142168,
-      "pages": 36,
+      "seconds": 1.5139052369995625,
+      "bytes": 6128565,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.3281040260008012,
-      "bytes": 876378,
-      "pages": 27,
-      "shards": 1
+      "seconds": 0.9614488920014992,
+      "bytes": 6537093,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.2359630550017755,
-      "bytes": 7977289,
-      "pages": 37,
+      "seconds": 1.2357967999996617,
+      "bytes": 11997860,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.4249079500004882,
-      "bytes": 10392834,
-      "pages": 36,
+      "seconds": 1.7381740189994161,
+      "bytes": 7938212,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.2200082600029418,
-      "bytes": 2504462,
+      "seconds": 1.2239695549978933,
+      "bytes": 5214335,
       "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.3331129750004038,
-      "bytes": 11935749,
+      "seconds": 1.2570640209996782,
+      "bytes": 6323177,
       "pages": 35,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.076318840001477,
-      "bytes": 1024576,
-      "pages": 34,
-      "shards": 1
+      "seconds": 1.1513970990017697,
+      "bytes": 2421895,
+      "pages": 31,
+      "shards": 2
     },
     {
-      "seconds": 1.0973008010005287,
-      "bytes": 11080849,
+      "seconds": 0.9733352420007577,
+      "bytes": 6126927,
+      "pages": 37,
+      "shards": 2
+    },
+    {
+      "seconds": 1.3508422129998507,
+      "bytes": 16469825,
       "pages": 33,
       "shards": 4
     },
     {
-      "seconds": 1.1580557140005112,
-      "bytes": 6155625,
-      "pages": 38,
+      "seconds": 1.220427556996583,
+      "bytes": 2558602,
+      "pages": 34,
       "shards": 2
     },
     {
-      "seconds": 0.8461053749997518,
-      "bytes": 1155308,
-      "pages": 32,
-      "shards": 1
+      "seconds": 1.1153945629994269,
+      "bytes": 6138686,
+      "pages": 33,
+      "shards": 2
     },
     {
-      "seconds": 1.2314064579986734,
-      "bytes": 7429916,
-      "pages": 31,
+      "seconds": 1.4000027559959562,
+      "bytes": 7623395,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.4280758529966988,
-      "bytes": 11440670,
-      "pages": 39,
+      "seconds": 1.0437422140021226,
+      "bytes": 887331,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3929757290025009,
+      "bytes": 11170736,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.364633949000563,
-      "bytes": 7290084,
-      "pages": 30,
+      "seconds": 1.5524120170011884,
+      "bytes": 8373554,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.0805251250021684,
-      "bytes": 6054722,
-      "pages": 36,
+      "seconds": 1.806914691995189,
+      "bytes": 5833325,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 2.882830210000975,
-      "bytes": 1047147,
-      "pages": 31,
+      "seconds": 1.1171102930020425,
+      "bytes": 1254155,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.6690692930023943,
-      "bytes": 11069831,
-      "pages": 40,
+      "seconds": 1.3684268549986882,
+      "bytes": 11656046,
+      "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 2.282538455998292,
-      "bytes": 6276829,
-      "pages": 36,
+      "seconds": 1.130367224002839,
+      "bytes": 6468621,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.4166184950008756,
-      "bytes": 8285024,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.1125211710022995,
+      "bytes": 2380554,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 1.271756786001788,
-      "bytes": 10114603,
-      "pages": 34,
+      "seconds": 1.3700068609978189,
+      "bytes": 12013278,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.2068859079990943,
-      "bytes": 6031296,
+      "seconds": 1.232138457999099,
+      "bytes": 9991874,
       "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.0317238650022773,
-      "bytes": 2334152,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 1.198212211002101,
-      "bytes": 5956669,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.182239486999606,
-      "bytes": 5356316,
-      "pages": 28,
+      "seconds": 1.27796818799834,
+      "bytes": 2573475,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.588657971999055,
-      "bytes": 13335657,
+      "seconds": 1.1166499749961076,
+      "bytes": 6085922,
       "pages": 33,
-      "shards": 4
-    },
-    {
-      "seconds": 0.8740047409992258,
-      "bytes": 947216,
-      "pages": 27,
-      "shards": 1
-    },
-    {
-      "seconds": 1.4317332190003071,
-      "bytes": 7796978,
-      "pages": 38,
-      "shards": 3
-    },
-    {
-      "seconds": 1.0188206910024746,
-      "bytes": 812811,
-      "pages": 23,
-      "shards": 1
-    },
-    {
-      "seconds": 1.1825939109985484,
-      "bytes": 10557366,
-      "pages": 46,
-      "shards": 3
-    },
-    {
-      "seconds": 1.6627466680001817,
-      "bytes": 7279837,
-      "pages": 30,
-      "shards": 3
+      "shards": 2
     },
     {
-      "seconds": 1.578546805998485,
-      "bytes": 6985484,
-      "pages": 38,
+      "seconds": 0.8856082540005445,
+      "bytes": 2405776,
+      "pages": 29,
       "shards": 2
     }
   ],

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fa840d52fec48122d7ba48ef7ead58d0b3533e834c3dec925aa67b3f13281da5
-size 721765951
+oid sha256:04389db27054d0deb18e21ea9848cc49ae3250a62872254d899b5d9fb8463946
+size 721760703