Qwen90M step 115000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 110000,
+  "step": 115000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "771a846dfcaa9ac2239f4c06cbb98a3687803f55e50cb2bc31108644214bc872",
+    "model.safetensors": "5bf76550c11679f0c60b72b0e0dcf8d8671863fe367e2ba0e086ace9b8c7a937",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "847a0836b0889eaaab0298c4b432da9e790a6c0dff6648e498d3e21db336447a",
+    "trainer.pt": "4343a4d314b627ad5a544efedc6122e5f21a7d7db1866f3e36e9dcac5761ca5c",
     "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783",
     "performance.json": "7e827680f2cb27108bc66b58bb4d6ede7b8c07fab9033e664cac4bed5acb1a98",
-    "shard_scheduler.json": "c064c98867fe8d56cbdc4767e22ae652e692137319af0d6e6bccc2225e28dd1b",
-    "coverage.json": "2ca6a281fc319220e745c11a5bb944816bba2680cf6eeb45cee83e9353622a10",
+    "shard_scheduler.json": "f5e86bb20ce677d559d3ccc0f4eb30f128b043a343171d0230f4ff2d770f4fa9",
+    "coverage.json": "19b38075c1f9099bca0be195b37ce43301f400314544de7407bf0ed838e03363",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 110000,
+  "step": 115000,
   "stats": {
-    "climbmix/blocks": 3960000,
-    "climbmix/targets": 6616550128,
-    "climbmix/input_tokens": 6616969711,
-    "ultra_qa/blocks": 1485000,
-    "ultra_qa/targets": 2422404971,
-    "ultra_qa/input_tokens": 2422409661,
-    "cortex/blocks": 1485000,
-    "cortex/targets": 56496990,
-    "cortex/input_tokens": 2360431026,
-    "ultra_style/blocks": 1485000,
-    "ultra_style/targets": 2348057065,
-    "ultra_style/input_tokens": 2348180586,
-    "openmath/blocks": 990000,
-    "openmath/targets": 1529649447,
-    "openmath/input_tokens": 1530159661,
-    "rewrite6/blocks": 495000,
-    "rewrite6/targets": 811489243,
-    "rewrite6/input_tokens": 812378858
+    "climbmix/blocks": 4140000,
+    "climbmix/targets": 6917414099,
+    "climbmix/input_tokens": 6917852847,
+    "ultra_qa/blocks": 1552500,
+    "ultra_qa/targets": 2532479019,
+    "ultra_qa/input_tokens": 2532483905,
+    "cortex/blocks": 1552500,
+    "cortex/targets": 59060192,
+    "cortex/input_tokens": 2467718909,
+    "ultra_style/blocks": 1552500,
+    "ultra_style/targets": 2455086974,
+    "ultra_style/input_tokens": 2455216177,
+    "openmath/blocks": 1035000,
+    "openmath/targets": 1599098904,
+    "openmath/input_tokens": 1599632106,
+    "rewrite6/blocks": 517500,
+    "rewrite6/targets": 848348966,
+    "rewrite6/input_tokens": 849279572
   },
   "sources": {
     "climbmix": {
-      "page": 542,
-      "row": 3742,
+      "page": 567,
+      "row": 2637,
       "start": {
-        "file": 26,
+        "file": 27,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 13,
+          "row_group": 30,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 2223774
+      "offset": 2048,
+      "documents": 2325069
     },
     "rewrite6": {
-      "page": 8,
-      "row": 2154,
+      "page": 22,
+      "row": 3071,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 32768
+          "row": 90112
         }
       },
       "epoch": 6,
       "offset": 0,
-      "documents": 1267346
+      "documents": 1325607
     },
     "openmath": {
-      "page": 339,
-      "row": 371,
+      "page": 354,
+      "row": 1854,
       "start": {
         "file": 3,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 78,
-          "row": 594
+          "row_group": 140,
+          "row": 34
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 1388915
+      "documents": 1451838
     },
     "ultra_qa": {
-      "page": 146,
-      "row": 2430,
+      "page": 153,
+      "row": 1077,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 1,
-          "row": 13902
+          "row": 42574
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 600446
+      "documents": 627765
     },
     "ultra_style": {
-      "page": 224,
-      "row": 65,
+      "page": 234,
+      "row": 827,
       "start": {
         "file": 1,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 2,
-          "row": 61881
+          "row_group": 3,
+          "row": 17255
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 917569
+      "offset": 2048,
+      "documents": 959291
     },
     "cortex": {
-      "page": 28796,
-      "row": 61,
+      "page": 30106,
+      "row": 48,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 3685949
+      "documents": 3853616
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 110000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "771a846dfcaa9ac2239f4c06cbb98a3687803f55e50cb2bc31108644214bc872", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "847a0836b0889eaaab0298c4b432da9e790a6c0dff6648e498d3e21db336447a", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "7e827680f2cb27108bc66b58bb4d6ede7b8c07fab9033e664cac4bed5acb1a98", "shard_scheduler.json": "c064c98867fe8d56cbdc4767e22ae652e692137319af0d6e6bccc2225e28dd1b", "coverage.json": "2ca6a281fc319220e745c11a5bb944816bba2680cf6eeb45cee83e9353622a10", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000110000"}
+{"step": 115000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "5bf76550c11679f0c60b72b0e0dcf8d8671863fe367e2ba0e086ace9b8c7a937", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "4343a4d314b627ad5a544efedc6122e5f21a7d7db1866f3e36e9dcac5761ca5c", "training_config.json": "95ba9e002790d11959d1191b36ed994bfa10a6da51c22c772e143bec517a9783", "performance.json": "7e827680f2cb27108bc66b58bb4d6ede7b8c07fab9033e664cac4bed5acb1a98", "shard_scheduler.json": "f5e86bb20ce677d559d3ccc0f4eb30f128b043a343171d0230f4ff2d770f4fa9", "coverage.json": "19b38075c1f9099bca0be195b37ce43301f400314544de7407bf0ed838e03363", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000115000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:771a846dfcaa9ac2239f4c06cbb98a3687803f55e50cb2bc31108644214bc872
+oid sha256:5bf76550c11679f0c60b72b0e0dcf8d8671863fe367e2ba0e086ace9b8c7a937
 size 360622744

shard_scheduler.json

@@ -22,8 +22,8 @@
       "tokens_per_second": 1062967.6168812835,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 75187.87582822153,
-      "required_pages_per_second_upper_estimate": 0.006517094514861157
+      "required_token_positions_per_second": 75399.42172020269,
+      "required_pages_per_second_upper_estimate": 0.006535430776619923
     },
     {
       "source": "rewrite6",
@@ -36,8 +36,8 @@
       "tokens_per_second": 1206672.263377472,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 9398.484478527691,
-      "required_pages_per_second_upper_estimate": 0.0037491959783499643
+      "required_token_positions_per_second": 9424.927715025337,
+      "required_pages_per_second_upper_estimate": 0.0037597445807504935
     },
     {
       "source": "openmath",
@@ -50,8 +50,8 @@
       "tokens_per_second": 410042.7581134458,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 18796.968957055382,
-      "required_pages_per_second_upper_estimate": 0.0043483479528598655
+      "required_token_positions_per_second": 18849.855430050673,
+      "required_pages_per_second_upper_estimate": 0.004360582307617191
     },
     {
       "source": "ultra_qa",
@@ -64,8 +64,8 @@
       "tokens_per_second": 2163486.5624068812,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28195.453435583073,
-      "required_pages_per_second_upper_estimate": 0.001716955184115474
+      "required_token_positions_per_second": 28274.78314507601,
+      "required_pages_per_second_upper_estimate": 0.0017217859472127746
     },
     {
       "source": "ultra_style",
@@ -78,8 +78,8 @@
       "tokens_per_second": 1546694.2737633225,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 1,
-      "required_token_positions_per_second": 28195.453435583073,
-      "required_pages_per_second_upper_estimate": 0.0021705640554817176
+      "required_token_positions_per_second": 28274.78314507601,
+      "required_pages_per_second_upper_estimate": 0.002176671076117175
     },
     {
       "source": "cortex",
@@ -92,8 +92,8 @@
       "tokens_per_second": 1015177.8395898147,
       "note": "Bounded fresh sample; full-page latency is an extrapolation.",
       "estimated_pages_for_buffer": 22,
-      "required_token_positions_per_second": 28195.453435583073,
-      "required_pages_per_second_upper_estimate": 0.3500931675575583
+      "required_token_positions_per_second": 28274.78314507601,
+      "required_pages_per_second_upper_estimate": 0.351078177050002
     }
   ],
   "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -158,410 +158,410 @@
     "prefetch_batches": 62,
     "prefetch_capacity_limit": 512,
     "batch_memory_estimate_bytes": 688946,
-    "producer_batch_seconds": 0.08555045376556336,
-    "producer_p95_seconds": 0.3201156809955137,
-    "gpu_update_seconds": 0.9805836271853611,
-    "producer_batches_per_second": 11.689008719232886,
-    "gpu_batches_per_second": 1.0198008331735775,
-    "producer_headroom": 11.462050568106696,
+    "producer_batch_seconds": 0.24759402542196085,
+    "producer_p95_seconds": 0.3587101380035165,
+    "gpu_update_seconds": 0.9778324331663296,
+    "producer_batches_per_second": 4.038869671009852,
+    "gpu_batches_per_second": 1.022670107967159,
+    "producer_headroom": 3.949337757645257,
     "sustainable": true,
-    "buffer_seconds": 60.79618488549239,
+    "buffer_seconds": 60.625610856312434,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 85,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 390327.88351984415,
-    "upload_bytes_per_second": 4067954.1604864304,
+    "generation_bytes_per_second": 416267.331035606,
+    "upload_bytes_per_second": 6126148.728639764,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 83.75527743526929,
+    "estimated_full_shards_per_hour": 89.32127903271804,
     "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": 1.0622286870020616,
-      "bytes": 2500050,
-      "pages": 30,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1363001400022767,
-      "bytes": 9937212,
+      "seconds": 1.1623649149987614,
+      "bytes": 11298342,
       "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 0.7940713860007236,
-      "bytes": 1067006,
-      "pages": 34,
-      "shards": 1
+      "seconds": 0.995095141002821,
+      "bytes": 6689673,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 0.9383102729989332,
-      "bytes": 7322897,
-      "pages": 29,
+      "seconds": 0.9397628990009252,
+      "bytes": 7665294,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.1287910210012342,
-      "bytes": 10964395,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.2407093820002046,
+      "bytes": 5771889,
+      "pages": 29,
+      "shards": 2
     },
     {
-      "seconds": 1.271768777998659,
-      "bytes": 8367852,
-      "pages": 34,
+      "seconds": 1.0350117630005116,
+      "bytes": 6519406,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 1.080213831999572,
-      "bytes": 6038268,
-      "pages": 31,
+      "seconds": 1.0306341840005189,
+      "bytes": 6113692,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.7658507999985886,
-      "bytes": 1224241,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.1684751539978606,
+      "bytes": 6202267,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 34.26547425300305,
-      "bytes": 7641108,
-      "pages": 29,
-      "shards": 4
+      "seconds": 1.093655648001004,
+      "bytes": 8128170,
+      "pages": 33,
+      "shards": 3
     },
     {
-      "seconds": 1.169081484000344,
-      "bytes": 10839288,
-      "pages": 47,
+      "seconds": 1.1689348799991421,
+      "bytes": 10022604,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.171269268001197,
-      "bytes": 2613392,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8061036489998514,
+      "bytes": 878150,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.150122408002062,
-      "bytes": 11429829,
-      "pages": 30,
+      "seconds": 1.0866164300023229,
+      "bytes": 7550162,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.3652685859997291,
-      "bytes": 10381816,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.2529126609988452,
+      "bytes": 5669267,
+      "pages": 30,
+      "shards": 2
     },
     {
-      "seconds": 1.0450683730014134,
-      "bytes": 2703655,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.9512047030002577,
+      "bytes": 7497853,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 0.9104454030020861,
-      "bytes": 6024195,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.1582007809993229,
+      "bytes": 12016223,
+      "pages": 34,
+      "shards": 3
     },
     {
-      "seconds": 1.20260026500182,
-      "bytes": 1126978,
-      "pages": 33,
+      "seconds": 0.9243304510018788,
+      "bytes": 1103405,
+      "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.128464265002549,
-      "bytes": 12438546,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.4641582209987973,
+      "bytes": 7254109,
+      "pages": 30,
+      "shards": 3
     },
     {
-      "seconds": 1.0890558419996523,
-      "bytes": 11718926,
-      "pages": 32,
+      "seconds": 1.1067068009979266,
+      "bytes": 10335265,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 0.9750485680015117,
-      "bytes": 2678604,
-      "pages": 36,
+      "seconds": 1.2287768099959067,
+      "bytes": 2576063,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 0.9471035349997692,
-      "bytes": 5896771,
-      "pages": 27,
+      "seconds": 1.1285306259960635,
+      "bytes": 5981374,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 0.970585907001805,
-      "bytes": 1087871,
-      "pages": 35,
+      "seconds": 0.8470438740041573,
+      "bytes": 878239,
+      "pages": 27,
       "shards": 1
     },
     {
-      "seconds": 1.300397458999214,
-      "bytes": 11506344,
-      "pages": 35,
-      "shards": 4
+      "seconds": 1.8046116949990392,
+      "bytes": 17147881,
+      "pages": 44,
+      "shards": 5
     },
     {
-      "seconds": 1.0356865150024532,
-      "bytes": 6075433,
-      "pages": 35,
-      "shards": 2
+      "seconds": 0.9106033309944905,
+      "bytes": 774779,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 0.8513431389983452,
-      "bytes": 631745,
-      "pages": 18,
-      "shards": 1
+      "seconds": 1.1979517419968033,
+      "bytes": 6268819,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.1167246349978086,
-      "bytes": 13461458,
+      "seconds": 1.2165552439983003,
+      "bytes": 11720658,
       "pages": 33,
       "shards": 4
     },
     {
-      "seconds": 1.146249319997878,
-      "bytes": 4617912,
-      "pages": 32,
-      "shards": 2
+      "seconds": 0.8457645430025877,
+      "bytes": 857794,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 0.8623779709996597,
-      "bytes": 6286843,
-      "pages": 36,
+      "seconds": 1.0460859910017462,
+      "bytes": 6362637,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.446577980001166,
-      "bytes": 2379048,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.0020113030041102,
+      "bytes": 8378438,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.0237194600013027,
-      "bytes": 6170294,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.1406177940007183,
+      "bytes": 9702248,
+      "pages": 32,
+      "shards": 3
     },
     {
-      "seconds": 1.2663461060001282,
-      "bytes": 11989707,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.1464950649969978,
+      "bytes": 7491850,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 0.9998438940019696,
-      "bytes": 6675780,
-      "pages": 28,
-      "shards": 2
+      "seconds": 0.8147378479989129,
+      "bytes": 902236,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.0791062770003919,
-      "bytes": 6176898,
-      "pages": 35,
+      "seconds": 1.2838300810035435,
+      "bytes": 5986754,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9380000849996577,
-      "bytes": 2661456,
+      "seconds": 1.19388542699744,
+      "bytes": 12323550,
       "pages": 37,
-      "shards": 2
+      "shards": 4
     },
     {
-      "seconds": 1.1787855490001675,
-      "bytes": 9821385,
-      "pages": 34,
-      "shards": 3
+      "seconds": 1.0270863580008154,
+      "bytes": 6920922,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.015233548998367,
-      "bytes": 5854223,
-      "pages": 28,
+      "seconds": 0.9876665679985308,
+      "bytes": 5790107,
+      "pages": 27,
       "shards": 2
     },
     {
-      "seconds": 0.8780484679991787,
-      "bytes": 2637899,
+      "seconds": 0.9691603890023543,
+      "bytes": 2643836,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1060435590006819,
-      "bytes": 11753109,
-      "pages": 33,
-      "shards": 3
+      "seconds": 1.1510760689998278,
+      "bytes": 5630242,
+      "pages": 21,
+      "shards": 2
     },
     {
-      "seconds": 1.0484200940009032,
-      "bytes": 6644677,
-      "pages": 34,
+      "seconds": 1.3745940619992325,
+      "bytes": 6263311,
+      "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9731116549992294,
-      "bytes": 6198048,
-      "pages": 36,
+      "seconds": 0.9876477170037106,
+      "bytes": 6238360,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2161200219998136,
-      "bytes": 5847296,
-      "pages": 28,
+      "seconds": 1.2104173959960463,
+      "bytes": 11923116,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.0819334130028437,
-      "bytes": 2586941,
-      "pages": 33,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1623649149987614,
-      "bytes": 11298342,
-      "pages": 37,
+      "seconds": 1.3382599670003401,
+      "bytes": 7199956,
+      "pages": 28,
       "shards": 3
     },
     {
-      "seconds": 0.995095141002821,
-      "bytes": 6689673,
-      "pages": 30,
+      "seconds": 1.2495986109934165,
+      "bytes": 6159261,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9397628990009252,
-      "bytes": 7665294,
+      "seconds": 1.4083330920038861,
+      "bytes": 7782602,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 1.2407093820002046,
-      "bytes": 5771889,
-      "pages": 29,
+      "seconds": 0.8252346610024688,
+      "bytes": 897645,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 1.2508048210002016,
+      "bytes": 6000035,
+      "pages": 32,
       "shards": 2
     },
     {
-      "seconds": 1.0350117630005116,
-      "bytes": 6519406,
-      "pages": 35,
-      "shards": 3
+      "seconds": 1.173013809006079,
+      "bytes": 12591906,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.0306341840005189,
-      "bytes": 6113692,
-      "pages": 36,
+      "seconds": 0.9398781409981893,
+      "bytes": 5766570,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.1684751539978606,
-      "bytes": 6202267,
+      "seconds": 1.1578597999978228,
+      "bytes": 6291400,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.093655648001004,
-      "bytes": 8128170,
-      "pages": 33,
+      "seconds": 1.1973399740018067,
+      "bytes": 6867540,
+      "pages": 31,
       "shards": 3
     },
     {
-      "seconds": 1.1689348799991421,
-      "bytes": 10022604,
-      "pages": 37,
-      "shards": 3
+      "seconds": 1.0004031040007249,
+      "bytes": 6252534,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 0.8061036489998514,
-      "bytes": 878150,
-      "pages": 26,
+      "seconds": 0.9013959430012619,
+      "bytes": 1172916,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.0866164300023229,
-      "bytes": 7550162,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.3720371540039196,
+      "bytes": 13095615,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "seconds": 1.2529126609988452,
-      "bytes": 5669267,
-      "pages": 30,
+      "seconds": 1.1362720050019561,
+      "bytes": 5854630,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9512047030002577,
-      "bytes": 7497853,
-      "pages": 35,
+      "seconds": 1.0005343860029825,
+      "bytes": 7274137,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.1582007809993229,
-      "bytes": 12016223,
-      "pages": 34,
-      "shards": 3
+      "seconds": 0.9837193970015505,
+      "bytes": 6134787,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.9243304510018788,
-      "bytes": 1103405,
+      "seconds": 0.8095773189997999,
+      "bytes": 1214722,
       "pages": 34,
       "shards": 1
     },
     {
-      "seconds": 1.4641582209987973,
-      "bytes": 7254109,
-      "pages": 30,
+      "seconds": 1.0524650669976836,
+      "bytes": 7528154,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.1067068009979266,
-      "bytes": 10335265,
-      "pages": 39,
+      "seconds": 1.1210896519987728,
+      "bytes": 9406921,
+      "pages": 24,
       "shards": 3
     },
     {
-      "seconds": 1.2287768099959067,
-      "bytes": 2576063,
-      "pages": 32,
+      "seconds": 0.883425399006228,
+      "bytes": 6847863,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1285306259960635,
-      "bytes": 5981374,
-      "pages": 32,
+      "seconds": 1.1994957349961624,
+      "bytes": 7913979,
+      "pages": 39,
+      "shards": 3
+    },
+    {
+      "seconds": 0.9844400470028631,
+      "bytes": 5585376,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 0.8470438740041573,
-      "bytes": 878239,
-      "pages": 27,
-      "shards": 1
+      "seconds": 1.2727246810027282,
+      "bytes": 7625766,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.8046116949990392,
-      "bytes": 17147881,
-      "pages": 44,
-      "shards": 5
+      "seconds": 1.004631062001863,
+      "bytes": 5973803,
+      "pages": 28,
+      "shards": 3
     },
     {
-      "seconds": 0.9106033309944905,
-      "bytes": 774779,
-      "pages": 26,
+      "seconds": 0.8451246749973507,
+      "bytes": 1134510,
+      "pages": 31,
       "shards": 1
     },
     {
-      "seconds": 1.1979517419968033,
-      "bytes": 6268819,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.538072485003795,
+      "bytes": 17347426,
+      "pages": 42,
+      "shards": 5
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:847a0836b0889eaaab0298c4b432da9e790a6c0dff6648e498d3e21db336447a
-size 721776063
+oid sha256:4343a4d314b627ad5a544efedc6122e5f21a7d7db1866f3e36e9dcac5761ca5c
+size 721765247