Qwen90M step 955000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 950000,
+  "step": 955000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "2e6800a411ea52928020ac842716553e33b1e7f874b99a56f4a6d2a6bbbab114",
+    "model.safetensors": "5efa97481adb165836eab13ec2e1d8440d32bc3f84314194d651cebe4c6e7d73",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "8ff2ea6bb34408b69ebbd7193690d1ba5ccb5272bc038889553875b472839158",
+    "trainer.pt": "af9b8f970077d4bd0f492458ce4f2991fcef79425d6050931c1082434f1da813",
     "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd",
     "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4",
-    "shard_scheduler.json": "b552fa7d9f27edf138aa08049c5810dc5f2939827c4512222fbf3c1ba1750c42",
-    "coverage.json": "b37ea7e378a52a99096460a590042a103f46dd59f031c3a7ade67958795c5436",
+    "shard_scheduler.json": "235d1326e974dcf3b57d55e9bb793f73668d9a67f58d4b3bce1cf5d4eec5e79a",
+    "coverage.json": "471d0cc109ff51b50b59b2ac675520cec412cadab32f02434f02213ec1811ae9",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 950000,
+  "step": 955000,
   "stats": {
-    "climbmix/blocks": 34200000,
-    "climbmix/targets": 57171948898,
-    "climbmix/input_tokens": 57175571843,
-    "ultra_qa/blocks": 12825000,
-    "ultra_qa/targets": 20922958426,
-    "ultra_qa/input_tokens": 20922998603,
-    "cortex/blocks": 12825000,
-    "cortex/targets": 487674525,
-    "cortex/input_tokens": 20386543238,
-    "ultra_style/blocks": 12825000,
-    "ultra_style/targets": 20285002197,
-    "ultra_style/input_tokens": 20286064763,
-    "openmath/blocks": 8550000,
-    "openmath/targets": 13211092129,
-    "openmath/input_tokens": 13215494259,
-    "rewrite6/blocks": 4275000,
-    "rewrite6/targets": 7008276799,
-    "rewrite6/input_tokens": 7015960317
+    "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
   },
   "sources": {
     "climbmix": {
-      "page": 4683,
-      "row": 3857,
+      "page": 4708,
+      "row": 2416,
       "start": {
-        "file": 226,
+        "file": 227,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 57,
+          "row_group": 75,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 8192,
-      "documents": 19185425
+      "offset": 0,
+      "documents": 19286384
     },
     "rewrite6": {
-      "page": 14,
-      "row": 2553,
+      "page": 28,
+      "row": 3208,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 57344
+          "row": 114688
         }
       },
       "epoch": 53,
-      "offset": 0,
-      "documents": 10946309
+      "offset": 2048,
+      "documents": 11004308
     },
     "openmath": {
-      "page": 2927,
-      "row": 1756,
+      "page": 2942,
+      "row": 3402,
       "start": {
         "file": 27,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 199,
-          "row": 446
+          "row_group": 260,
+          "row": 886
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 11990748
+      "offset": 2048,
+      "documents": 12053834
     },
     "ultra_qa": {
-      "page": 1265,
-      "row": 2680,
+      "page": 1272,
+      "row": 1325,
       "start": {
-        "file": 9,
+        "file": 10,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 7,
-          "row": 52881
+          "row_group": 0,
+          "row": 13836
         }
       },
       "epoch": 0,
-      "offset": 4096,
-      "documents": 5184120
+      "offset": 2048,
+      "documents": 5211437
     },
     "ultra_style": {
-      "page": 1931,
-      "row": 1294,
+      "page": 1941,
+      "row": 2726,
       "start": {
         "file": 11,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 35497
+          "row": 76457
         }
       },
       "epoch": 0,
-      "offset": 0,
-      "documents": 7910670
+      "offset": 4096,
+      "documents": 7953062
     },
     "cortex": {
-      "page": 248725,
-      "row": 90,
+      "page": 250033,
+      "row": 97,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 31836890
+      "documents": 32004321
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 950000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "2e6800a411ea52928020ac842716553e33b1e7f874b99a56f4a6d2a6bbbab114", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "8ff2ea6bb34408b69ebbd7193690d1ba5ccb5272bc038889553875b472839158", "training_config.json": "4e8f8b2fc7193aabdbca12082e7c7cc3a0ed080bd5ff16e060cc02e10ffc53dd", "performance.json": "a7562f2130d78c752d5494512bcbf6d4c62627eee4ff6943c08f7615044b2de4", "shard_scheduler.json": "b552fa7d9f27edf138aa08049c5810dc5f2939827c4512222fbf3c1ba1750c42", "coverage.json": "b37ea7e378a52a99096460a590042a103f46dd59f031c3a7ade67958795c5436", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "052aedaef69b550457da29474e17df28fa635c99594a7b9c0881f3ac3827000e", "folder": "step-000000950000"}
+{"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"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2e6800a411ea52928020ac842716553e33b1e7f874b99a56f4a6d2a6bbbab114
+oid sha256:5efa97481adb165836eab13ec2e1d8440d32bc3f84314194d651cebe4c6e7d73
 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": 76668.71420286849,
-      "required_pages_per_second_upper_estimate": 0.005304119544480668
+      "required_token_positions_per_second": 76445.90440437887,
+      "required_pages_per_second_upper_estimate": 0.005288705045631197
     },
     {
       "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": 9583.589275358561,
-      "required_pages_per_second_upper_estimate": 0.004012448701491235
+      "required_token_positions_per_second": 9555.73805054736,
+      "required_pages_per_second_upper_estimate": 0.0040007879752624945
     },
     {
       "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": 19167.178550717123,
-      "required_pages_per_second_upper_estimate": 0.003992017819624442
+      "required_token_positions_per_second": 19111.47610109472,
+      "required_pages_per_second_upper_estimate": 0.003980416468340475
     },
     {
       "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": 28750.767826075684,
-      "required_pages_per_second_upper_estimate": 0.0017235646312926524
+      "required_token_positions_per_second": 28667.214151642078,
+      "required_pages_per_second_upper_estimate": 0.0017185557160894315
     },
     {
       "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": 28750.767826075684,
-      "required_pages_per_second_upper_estimate": 0.0022352457372145557
+      "required_token_positions_per_second": 28667.214151642078,
+      "required_pages_per_second_upper_estimate": 0.002228749806541117
     },
     {
       "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": 28750.767826075684,
-      "required_pages_per_second_upper_estimate": 0.3687839795035426
+      "required_token_positions_per_second": 28667.214151642078,
+      "required_pages_per_second_upper_estimate": 0.36771224268085423
     }
   ],
   "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": 0.10526186014067207,
-    "producer_p95_seconds": 0.27886092599874246,
-    "gpu_update_seconds": 0.9616438826000493,
-    "producer_batches_per_second": 9.500117123748325,
-    "gpu_batches_per_second": 1.039885989079705,
-    "producer_headroom": 9.135729516036552,
+    "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,
     "sustainable": true,
-    "buffer_seconds": 60.583564603803104,
+    "buffer_seconds": 60.76014190936747,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 82,
+    "upload_pages_per_commit": 79,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 357854.62686666014,
-    "upload_bytes_per_second": 4807959.195163555,
+    "generation_bytes_per_second": 340180.5052731563,
+    "upload_bytes_per_second": 5134244.0499124015,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 76.78727249622204,
+    "estimated_full_shards_per_hour": 72.99481743474976,
     "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": 1.4908248419997108,
-      "bytes": 6812358,
+      "seconds": 1.1959222149998823,
+      "bytes": 7412774,
       "pages": 30,
-      "shards": 2
-    },
-    {
-      "seconds": 1.521788887999719,
-      "bytes": 7493690,
-      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.3138526910006476,
-      "bytes": 6032843,
-      "pages": 36,
+      "seconds": 1.488441301000421,
+      "bytes": 6011041,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2020050520004588,
-      "bytes": 6391888,
-      "pages": 29,
-      "shards": 3
-    },
-    {
-      "seconds": 1.6235010039999906,
-      "bytes": 6476306,
-      "pages": 37,
+      "seconds": 1.0648157450013969,
+      "bytes": 668578,
+      "pages": 20,
       "shards": 2
     },
     {
-      "seconds": 3.9122833619985613,
-      "bytes": 6289054,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.9570103289988765,
+      "bytes": 18490744,
+      "pages": 47,
+      "shards": 5
     },
     {
-      "seconds": 1.8021375259995693,
-      "bytes": 8272426,
+      "seconds": 0.8552882639996824,
+      "bytes": 1175820,
       "pages": 35,
-      "shards": 3
+      "shards": 1
     },
     {
-      "seconds": 1.257975387001352,
-      "bytes": 10890364,
+      "seconds": 1.1064946840015182,
+      "bytes": 7293995,
       "pages": 29,
       "shards": 3
     },
     {
-      "seconds": 1.6343959760015423,
-      "bytes": 2597952,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.3388132420004695,
-      "bytes": 5864353,
-      "pages": 35,
+      "seconds": 1.736566771996877,
+      "bytes": 6191401,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 1.5614973880001344,
-      "bytes": 5966551,
-      "pages": 30,
+      "seconds": 1.1331273229989165,
+      "bytes": 4688984,
+      "pages": 31,
       "shards": 2
     },
     {
-      "seconds": 3.168707209999411,
-      "bytes": 6305368,
-      "pages": 39,
-      "shards": 4
-    },
-    {
-      "seconds": 1.9134112159990764,
-      "bytes": 11699377,
-      "pages": 28,
+      "seconds": 1.3685046969985706,
+      "bytes": 7267431,
+      "pages": 27,
       "shards": 3
     },
     {
-      "seconds": 0.9410961429966846,
-      "bytes": 1236103,
-      "pages": 35,
-      "shards": 1
+      "seconds": 1.4039977110005566,
+      "bytes": 12182760,
+      "pages": 48,
+      "shards": 3
     },
     {
-      "seconds": 1.262457610002457,
-      "bytes": 7619870,
-      "pages": 38,
-      "shards": 3
+      "seconds": 1.015954297999997,
+      "bytes": 2619113,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.496031877999485,
-      "bytes": 10106377,
-      "pages": 30,
+      "seconds": 1.3503473790005955,
+      "bytes": 10477535,
+      "pages": 34,
       "shards": 3
     },
     {
-      "seconds": 1.0667447009982425,
-      "bytes": 2533809,
-      "pages": 37,
-      "shards": 2
+      "seconds": 0.9818654609989608,
+      "bytes": 963058,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.2533954940008698,
-      "bytes": 6219875,
-      "pages": 36,
-      "shards": 2
+      "seconds": 1.1789819430014177,
+      "bytes": 7759537,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.2242291910006315,
-      "bytes": 6745128,
-      "pages": 29,
-      "shards": 2
+      "seconds": 1.2661478619993431,
+      "bytes": 11756535,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.3833632300011232,
-      "bytes": 11874470,
-      "pages": 36,
-      "shards": 4
+      "seconds": 1.2254978609998943,
+      "bytes": 7186634,
+      "pages": 35,
+      "shards": 3
     },
     {
-      "seconds": 1.439283288000297,
-      "bytes": 6018491,
-      "pages": 36,
+      "seconds": 1.2788624959976005,
+      "bytes": 6210071,
+      "pages": 38,
       "shards": 2
     },
     {
-      "seconds": 0.925738228001137,
-      "bytes": 863516,
-      "pages": 29,
+      "seconds": 1.412427707000461,
+      "bytes": 887856,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 1.338383825001074,
-      "bytes": 7489532,
-      "pages": 36,
+      "seconds": 1.2455697520017566,
+      "bytes": 7910132,
+      "pages": 40,
       "shards": 3
     },
     {
-      "seconds": 1.0948180369996408,
-      "bytes": 10021223,
-      "pages": 34,
+      "seconds": 1.3195631069975207,
+      "bytes": 10098822,
+      "pages": 36,
       "shards": 3
     },
     {
-      "seconds": 1.3469298159980099,
-      "bytes": 8271893,
-      "pages": 34,
+      "seconds": 1.6743710149967228,
+      "bytes": 8331470,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.2550481799989939,
-      "bytes": 6009424,
-      "pages": 28,
+      "seconds": 1.060381152001355,
+      "bytes": 6181629,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 0.9456898729986278,
-      "bytes": 975770,
-      "pages": 34,
-      "shards": 1
-    },
-    {
-      "seconds": 1.7259603879974748,
-      "bytes": 12302933,
+      "seconds": 1.3323454749988741,
+      "bytes": 6231984,
       "pages": 37,
-      "shards": 4
+      "shards": 2
     },
     {
-      "seconds": 1.298020391001046,
-      "bytes": 6312632,
-      "pages": 38,
-      "shards": 2
+      "seconds": 1.3298568760001217,
+      "bytes": 7493698,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 1.6241299210014404,
-      "bytes": 2390001,
+      "seconds": 1.2196332230014377,
+      "bytes": 5954986,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 1.5602491789977648,
-      "bytes": 11841247,
-      "pages": 32,
+      "seconds": 1.0563156829994114,
+      "bytes": 932363,
+      "pages": 27,
+      "shards": 1
+    },
+    {
+      "seconds": 1.3606180970018613,
+      "bytes": 7382890,
+      "pages": 28,
       "shards": 3
     },
     {
-      "seconds": 1.3097548330006248,
-      "bytes": 6099339,
-      "pages": 40,
-      "shards": 2
+      "seconds": 1.5418479689978994,
+      "bytes": 10959545,
+      "pages": 43,
+      "shards": 3
     },
     {
-      "seconds": 1.2393856140006392,
-      "bytes": 7638444,
-      "pages": 39,
+      "seconds": 1.2728104929992696,
+      "bytes": 7484326,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.58254548900004,
-      "bytes": 5975386,
-      "pages": 28,
+      "seconds": 1.4130519919999642,
+      "bytes": 6156286,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 0.9414393419974658,
-      "bytes": 1125185,
-      "pages": 34,
+      "seconds": 0.9332762750018446,
+      "bytes": 1158586,
+      "pages": 35,
       "shards": 1
     },
     {
-      "seconds": 1.356874795001204,
-      "bytes": 11516865,
+      "seconds": 1.577455007001845,
+      "bytes": 11622766,
       "pages": 35,
       "shards": 4
     },
     {
-      "seconds": 0.9571627120021731,
-      "bytes": 762513,
-      "pages": 23,
+      "seconds": 1.0520612039981643,
+      "bytes": 811209,
+      "pages": 24,
       "shards": 1
     },
     {
-      "seconds": 1.4426862469990738,
-      "bytes": 13691203,
-      "pages": 49,
+      "seconds": 1.4460338999997475,
+      "bytes": 11954296,
+      "pages": 39,
       "shards": 4
     },
     {
-      "seconds": 2.0410720630025025,
-      "bytes": 6120614,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.3092962549999356,
+      "bytes": 7432740,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.370815718000813,
-      "bytes": 5076360,
-      "pages": 34,
+      "seconds": 1.1428929149988107,
+      "bytes": 5656732,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1959222149998823,
-      "bytes": 7412774,
-      "pages": 30,
+      "seconds": 1.3837304669978039,
+      "bytes": 7550021,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.488441301000421,
-      "bytes": 6011041,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.4462293939977826,
+      "bytes": 810671,
+      "pages": 26,
+      "shards": 1
     },
     {
-      "seconds": 1.0648157450013969,
-      "bytes": 668578,
-      "pages": 20,
-      "shards": 2
+      "seconds": 1.2543598109987215,
+      "bytes": 12226155,
+      "pages": 42,
+      "shards": 3
     },
     {
-      "seconds": 1.9570103289988765,
-      "bytes": 18490744,
-      "pages": 47,
-      "shards": 5
+      "seconds": 1.268981666999025,
+      "bytes": 11497273,
+      "pages": 33,
+      "shards": 4
     },
     {
-      "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": 1
+    },
+    {
+      "seconds": 1.3873778990018764,
+      "bytes": 7922754,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.060381152001355,
-      "bytes": 6181629,
-      "pages": 30,
-      "shards": 2
+      "seconds": 1.5672195150000334,
+      "bytes": 991660,
+      "pages": 27,
+      "shards": 1
     },
     {
-      "seconds": 1.3323454749988741,
-      "bytes": 6231984,
-      "pages": 37,
-      "shards": 2
+      "seconds": 1.7213281089971133,
+      "bytes": 11058082,
+      "pages": 32,
+      "shards": 4
     },
     {
-      "seconds": 1.3298568760001217,
-      "bytes": 7493698,
-      "pages": 36,
-      "shards": 3
+      "seconds": 1.1525909759984643,
+      "bytes": 12461215,
+      "pages": 46,
+      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8ff2ea6bb34408b69ebbd7193690d1ba5ccb5272bc038889553875b472839158
-size 721758527
+oid sha256:af9b8f970077d4bd0f492458ce4f2991fcef79425d6050931c1082434f1da813
+size 721766911