Qwen90M step 460000; complete resumable checkpoint

complete.json

@@ -1,18 +1,18 @@
 {
-  "step": 455000,
+  "step": 460000,
   "files": {
     "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
     "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
-    "model.safetensors": "4e0571b196c03b2c4a354365ed7cfb6e857d56cf683df316f0f243fc3c3c4a92",
+    "model.safetensors": "e6df0b7486a13c4147338503b6e10d212d415f061b32918c019208df2361f792",
     "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
     "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
     "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
     "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
-    "trainer.pt": "ed5c494fe01b44806d348a5f83f4d51c818b578b1ae84966900fdecfbb1e84ce",
+    "trainer.pt": "f506a4eb49b59d5bfe154e01f4234e6f537368de7285d907d56b6e46d403a484",
     "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6",
     "performance.json": "58f8b6f79b191aa11b2bedb4594ca2897b6fdad2419ea7c9210688ac8022e7cb",
-    "shard_scheduler.json": "140235f913903bf3fca7092448b8c6b8f621ac3103af86fcdc689e03b692a948",
-    "coverage.json": "55cfee9475ccaed1654f1769d3d57fbdf4c9d032a2cd8b8c220c97dac7f99b83",
+    "shard_scheduler.json": "7aaac141beba6f945d5ac3dbede22069e5e37ddff308a0bbb94744199f0c5c6f",
+    "coverage.json": "74bb28e87fd46e4a0566d1ba75ad97ced9b1c5b7545ee80b386415b6215c402a",
     "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
   },
   "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"

coverage.json

@@ -1,107 +1,107 @@
 {
-  "step": 455000,
+  "step": 460000,
   "stats": {
-    "climbmix/blocks": 16380000,
-    "climbmix/targets": 27381108999,
-    "climbmix/input_tokens": 27382843118,
-    "ultra_qa/blocks": 6142500,
-    "ultra_qa/targets": 10020265694,
-    "ultra_qa/input_tokens": 10020284934,
-    "cortex/blocks": 6142500,
-    "cortex/targets": 233621534,
-    "cortex/input_tokens": 9764332089,
-    "ultra_style/blocks": 6142500,
-    "ultra_style/targets": 9715613685,
-    "ultra_style/input_tokens": 9716122899,
-    "openmath/blocks": 4095000,
-    "openmath/targets": 6327662730,
-    "openmath/input_tokens": 6329771233,
-    "rewrite6/blocks": 2047500,
-    "rewrite6/targets": 3356578938,
-    "rewrite6/input_tokens": 3360259205
+    "climbmix/blocks": 16560000,
+    "climbmix/targets": 27681620870,
+    "climbmix/input_tokens": 27683374095,
+    "ultra_qa/blocks": 6210000,
+    "ultra_qa/targets": 10130524498,
+    "ultra_qa/input_tokens": 10130543970,
+    "cortex/blocks": 6210000,
+    "cortex/targets": 236187228,
+    "cortex/input_tokens": 9871713742,
+    "ultra_style/blocks": 6210000,
+    "ultra_style/targets": 9822416214,
+    "ultra_style/input_tokens": 9822930973,
+    "openmath/blocks": 4140000,
+    "openmath/targets": 6397270462,
+    "openmath/input_tokens": 6399402063,
+    "rewrite6/blocks": 2070000,
+    "rewrite6/targets": 3393477095,
+    "rewrite6/input_tokens": 3397197651
   },
   "sources": {
     "climbmix": {
-      "page": 2243,
-      "row": 688,
+      "page": 2267,
+      "row": 3462,
       "start": {
-        "file": 108,
+        "file": 109,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 47,
+          "row_group": 60,
           "row": 1024
         }
       },
       "epoch": 0,
-      "offset": 6144,
-      "documents": 9188016
+      "offset": 2048,
+      "documents": 9289094
     },
     "rewrite6": {
-      "page": 26,
-      "row": 1443,
+      "page": 40,
+      "row": 1569,
       "start": {
         "file": 0,
         "reader": {
           "format": "jsonl-row-v1",
-          "row": 106496
+          "row": 163840
         }
       },
       "epoch": 25,
-      "offset": 0,
-      "documents": 5243039
+      "offset": 2048,
+      "documents": 5300509
     },
     "openmath": {
-      "page": 1402,
-      "row": 221,
+      "page": 1417,
+      "row": 1939,
       "start": {
         "file": 13,
         "reader": {
           "format": "parquet-row-v1",
-          "row_group": 66,
-          "row": 142
+          "row_group": 127,
+          "row": 582
         }
       },
       "epoch": 0,
       "offset": 0,
-      "documents": 5742813
+      "documents": 5805971
     },
     "ultra_qa": {
-      "page": 606,
-      "row": 864,
+      "page": 612,
+      "row": 3622,
       "start": {
         "file": 4,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 6,
-          "row": 16008
+          "row": 40584
         }
       },
       "epoch": 0,
-      "offset": 2048,
-      "documents": 2483040
+      "offset": 4096,
+      "documents": 2510374
     },
     "ultra_style": {
-      "page": 925,
-      "row": 539,
+      "page": 935,
+      "row": 1084,
       "start": {
         "file": 5,
         "reader": {
           "format": "parquet-row-v1",
           "row_group": 4,
-          "row": 23497
+          "row": 64457
         }
       },
       "epoch": 0,
       "offset": 2048,
-      "documents": 3789339
+      "documents": 3830844
     },
     "cortex": {
-      "page": 119135,
-      "row": 53,
+      "page": 120444,
+      "row": 91,
       "start": null,
       "epoch": 0,
       "offset": 0,
-      "documents": 15249333
+      "documents": 15416923
     }
   }
 }

latest.json

@@ -1 +1 @@
-{"step": 455000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "4e0571b196c03b2c4a354365ed7cfb6e857d56cf683df316f0f243fc3c3c4a92", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "ed5c494fe01b44806d348a5f83f4d51c818b578b1ae84966900fdecfbb1e84ce", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "58f8b6f79b191aa11b2bedb4594ca2897b6fdad2419ea7c9210688ac8022e7cb", "shard_scheduler.json": "140235f913903bf3fca7092448b8c6b8f621ac3103af86fcdc689e03b692a948", "coverage.json": "55cfee9475ccaed1654f1769d3d57fbdf4c9d032a2cd8b8c220c97dac7f99b83", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000455000"}
+{"step": 460000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "e6df0b7486a13c4147338503b6e10d212d415f061b32918c019208df2361f792", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "f506a4eb49b59d5bfe154e01f4234e6f537368de7285d907d56b6e46d403a484", "training_config.json": "c8bdf052510c2d9092e5a6824b119dd6584dc7697f8c4a8e9b7ecf3057c5bbb6", "performance.json": "58f8b6f79b191aa11b2bedb4594ca2897b6fdad2419ea7c9210688ac8022e7cb", "shard_scheduler.json": "7aaac141beba6f945d5ac3dbede22069e5e37ddff308a0bbb94744199f0c5c6f", "coverage.json": "74bb28e87fd46e4a0566d1ba75ad97ced9b1c5b7545ee80b386415b6215c402a", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000460000"}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4e0571b196c03b2c4a354365ed7cfb6e857d56cf683df316f0f243fc3c3c4a92
+oid sha256:e6df0b7486a13c4147338503b6e10d212d415f061b32918c019208df2361f792
 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": 76550.43485735178,
-      "required_pages_per_second_upper_estimate": 0.005295936704911929
+      "required_token_positions_per_second": 77107.71522175097,
+      "required_pages_per_second_upper_estimate": 0.0053344906535896615
     },
     {
       "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": 9568.804357168972,
-      "required_pages_per_second_upper_estimate": 0.004006258564989454
+      "required_token_positions_per_second": 9638.464402718871,
+      "required_pages_per_second_upper_estimate": 0.00403542377139403
     },
     {
       "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": 19137.608714337945,
-      "required_pages_per_second_upper_estimate": 0.003985859202515684
+      "required_token_positions_per_second": 19276.928805437743,
+      "required_pages_per_second_upper_estimate": 0.004014875903373896
     },
     {
       "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": 28706.41307150692,
-      "required_pages_per_second_upper_estimate": 0.0017209056314820437
+      "required_token_positions_per_second": 28915.393208156613,
+      "required_pages_per_second_upper_estimate": 0.0017334336715800003
     },
     {
       "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": 28706.41307150692,
-      "required_pages_per_second_upper_estimate": 0.0022317973501427813
+      "required_token_positions_per_second": 28915.393208156613,
+      "required_pages_per_second_upper_estimate": 0.0022480446365608186
     },
     {
       "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": 28706.41307150692,
-      "required_pages_per_second_upper_estimate": 0.36821504433635943
+      "required_token_positions_per_second": 28915.393208156613,
+      "required_pages_per_second_upper_estimate": 0.37089561714391317
     }
   ],
   "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": 704566,
-    "producer_batch_seconds": 0.2695831617967599,
-    "producer_p95_seconds": 0.24042049499985296,
-    "gpu_update_seconds": 0.9631297344997287,
-    "producer_batches_per_second": 3.7094304901502158,
-    "gpu_batches_per_second": 1.0382817227830916,
-    "producer_headroom": 3.572662803123576,
+    "producer_batch_seconds": 0.40328230902341033,
+    "producer_p95_seconds": 0.21793054399677203,
+    "gpu_update_seconds": 0.956168909790267,
+    "producer_batches_per_second": 2.479652535271391,
+    "gpu_batches_per_second": 1.0458403214755718,
+    "producer_headroom": 2.3709666613091174,
     "sustainable": true,
-    "buffer_seconds": 60.67717327348291,
+    "buffer_seconds": 60.238641316786826,
     "buffer_drain_seconds": null,
     "shard_target_mib": 16,
-    "upload_pages_per_commit": 79,
+    "upload_pages_per_commit": 83,
     "max_pending_pages": 4096,
-    "generation_bytes_per_second": 506639.92217322264,
-    "upload_bytes_per_second": 5145170.27868784,
+    "generation_bytes_per_second": 290697.8459376837,
+    "upload_bytes_per_second": 5108075.817065287,
     "upload_sustainable": true,
     "logical_page_rows": 4096,
     "logical_cortex_page_rows": 128,
-    "estimated_full_shards_per_hour": 108.71313332459935,
+    "estimated_full_shards_per_hour": 62.376990638712726,
     "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": 0.9096585289998984,
-      "bytes": 2702720,
-      "pages": 37,
-      "shards": 2
-    },
-    {
-      "seconds": 1.212294140999802,
-      "bytes": 5935208,
-      "pages": 36,
-      "shards": 2
-    },
-    {
-      "seconds": 1.1118033149996336,
-      "bytes": 6738611,
-      "pages": 31,
-      "shards": 2
-    },
-    {
-      "seconds": 1.2066322789996775,
-      "bytes": 7050449,
-      "pages": 25,
-      "shards": 3
-    },
-    {
-      "seconds": 1.1100294799998665,
-      "bytes": 10655210,
-      "pages": 43,
-      "shards": 3
+      "seconds": 1.142588306000107,
+      "bytes": 12693549,
+      "pages": 49,
+      "shards": 4
     },
     {
-      "seconds": 0.9483619619995807,
-      "bytes": 1257250,
-      "pages": 35,
+      "seconds": 0.9721095380009501,
+      "bytes": 837498,
+      "pages": 26,
       "shards": 1
     },
     {
-      "seconds": 0.9075711420000516,
-      "bytes": 7428003,
-      "pages": 38,
-      "shards": 3
+      "seconds": 0.9322610279996297,
+      "bytes": 6251371,
+      "pages": 36,
+      "shards": 2
     },
     {
-      "seconds": 1.278834224000093,
-      "bytes": 9834674,
-      "pages": 30,
+      "seconds": 1.1527972080002655,
+      "bytes": 6634416,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.1737758689996554,
-      "bytes": 8190004,
+      "seconds": 1.1540112700004101,
+      "bytes": 11807839,
       "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.150645027999417,
-      "bytes": 6104360,
-      "pages": 37,
+      "seconds": 0.8518780800004606,
+      "bytes": 2360936,
+      "pages": 30,
       "shards": 2
     },
     {
-      "seconds": 1.1603049650002504,
-      "bytes": 1160533,
+      "seconds": 1.2458887120010331,
+      "bytes": 6052831,
       "pages": 35,
-      "shards": 1
+      "shards": 2
     },
     {
-      "seconds": 1.0780582049992518,
-      "bytes": 12237678,
-      "pages": 34,
-      "shards": 4
+      "seconds": 1.0937011059995712,
+      "bytes": 11191807,
+      "pages": 40,
+      "shards": 3
     },
     {
-      "seconds": 0.9231663379996462,
-      "bytes": 6121395,
-      "pages": 36,
+      "seconds": 1.0167957749999914,
+      "bytes": 2251564,
+      "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.8000806340005511,
-      "bytes": 2307263,
-      "pages": 28,
+      "seconds": 0.9138518070012651,
+      "bytes": 6438645,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.328747977000603,
-      "bytes": 11901603,
-      "pages": 40,
+      "seconds": 1.2295570830010547,
+      "bytes": 11846540,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.0515727450001577,
-      "bytes": 10443094,
-      "pages": 35,
+      "seconds": 0.9956274190008116,
+      "bytes": 6486340,
+      "pages": 41,
       "shards": 3
     },
     {
-      "seconds": 0.9829580149998947,
-      "bytes": 2655368,
+      "seconds": 1.1072240289995534,
+      "bytes": 6495721,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.023656208999455,
-      "bytes": 5985547,
-      "pages": 27,
+      "seconds": 0.8532074099985039,
+      "bytes": 2480365,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.1010237709997455,
-      "bytes": 2681477,
+      "seconds": 0.959426125000391,
+      "bytes": 5985197,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0707301010006631,
-      "bytes": 10039688,
-      "pages": 34,
+      "seconds": 1.0452366619992972,
+      "bytes": 10510253,
+      "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.0178890370007139,
-      "bytes": 11702648,
-      "pages": 30,
+      "seconds": 1.1982196040007693,
+      "bytes": 8227634,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.0858605960002023,
-      "bytes": 2526198,
+      "seconds": 1.881202160999237,
+      "bytes": 6307877,
       "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.2877745800014964,
-      "bytes": 5791906,
-      "pages": 28,
-      "shards": 2
-    },
-    {
-      "seconds": 0.9569596109995473,
-      "bytes": 5245257,
-      "pages": 33,
-      "shards": 2
+      "seconds": 1.1178816689989617,
+      "bytes": 7521323,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 1.285435239000435,
-      "bytes": 7473995,
-      "pages": 31,
+      "seconds": 1.323959568000646,
+      "bytes": 5092035,
+      "pages": 32,
       "shards": 3
     },
     {
-      "seconds": 1.144911826999305,
-      "bytes": 6031558,
+      "seconds": 1.185996062000413,
+      "bytes": 6230664,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1537458799994056,
-      "bytes": 6763666,
-      "pages": 30,
+      "seconds": 0.9238252899995132,
+      "bytes": 2424834,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.232765078999364,
-      "bytes": 12254761,
+      "seconds": 1.1364503580007295,
+      "bytes": 11698392,
+      "pages": 34,
+      "shards": 3
+    },
+    {
+      "seconds": 1.0955314789989643,
+      "bytes": 11552086,
       "pages": 37,
       "shards": 4
     },
     {
-      "seconds": 1.0962526939983945,
-      "bytes": 1063767,
-      "pages": 31,
+      "seconds": 0.7776427819990204,
+      "bytes": 236866,
+      "pages": 8,
       "shards": 1
     },
     {
-      "seconds": 1.2301429309991363,
-      "bytes": 7644122,
-      "pages": 35,
-      "shards": 4
+      "seconds": 0.9514035880001757,
+      "bytes": 6117975,
+      "pages": 37,
+      "shards": 2
     },
     {
-      "seconds": 1.0025934540008166,
-      "bytes": 6106895,
-      "pages": 36,
-      "shards": 2
+      "seconds": 0.8122407450009632,
+      "bytes": 1177705,
+      "pages": 35,
+      "shards": 1
     },
     {
-      "seconds": 1.0866103590014973,
-      "bytes": 4773621,
-      "pages": 30,
-      "shards": 2
+      "seconds": 0.9919227340014913,
+      "bytes": 7505854,
+      "pages": 31,
+      "shards": 3
     },
     {
-      "seconds": 1.0814739849993202,
-      "bytes": 2089687,
-      "pages": 19,
+      "seconds": 1.4834884449992387,
+      "bytes": 5324170,
+      "pages": 33,
       "shards": 2
     },
     {
-      "seconds": 1.1024076169996988,
-      "bytes": 12461960,
-      "pages": 47,
-      "shards": 3
+      "seconds": 1.2237625110010413,
+      "bytes": 13342349,
+      "pages": 36,
+      "shards": 4
     },
     {
-      "seconds": 1.0951313269997627,
-      "bytes": 6163611,
+      "seconds": 1.0075590680007736,
+      "bytes": 6216793,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.1898623689994565,
-      "bytes": 6904260,
-      "pages": 25,
-      "shards": 3
+      "seconds": 0.8775070329993468,
+      "bytes": 1128371,
+      "pages": 32,
+      "shards": 1
     },
     {
-      "seconds": 1.1612172679997457,
-      "bytes": 6026166,
-      "pages": 34,
-      "shards": 2
+      "seconds": 1.7969425399987813,
+      "bytes": 12329538,
+      "pages": 40,
+      "shards": 4
     },
     {
-      "seconds": 1.3405283600004623,
-      "bytes": 7840338,
+      "seconds": 0.9383213960009016,
+      "bytes": 933737,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 3.437801178999507,
+      "bytes": 7738259,
       "pages": 38,
       "shards": 3
     },
     {
-      "seconds": 0.9687580539994087,
-      "bytes": 835437,
-      "pages": 25,
-      "shards": 1
+      "seconds": 1.1072633740004676,
+      "bytes": 11735446,
+      "pages": 36,
+      "shards": 3
     },
     {
-      "seconds": 14.438800792999245,
-      "bytes": 11778517,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.0690352890014765,
+      "bytes": 5156190,
+      "pages": 34,
+      "shards": 2
     },
     {
-      "seconds": 1.142588306000107,
-      "bytes": 12693549,
-      "pages": 49,
-      "shards": 4
+      "seconds": 1.2706896670024435,
+      "bytes": 7639609,
+      "pages": 39,
+      "shards": 3
     },
     {
-      "seconds": 0.9721095380009501,
-      "bytes": 837498,
-      "pages": 26,
-      "shards": 1
+      "seconds": 1.257235208999191,
+      "bytes": 5936765,
+      "pages": 28,
+      "shards": 2
     },
     {
-      "seconds": 0.9322610279996297,
-      "bytes": 6251371,
-      "pages": 36,
+      "seconds": 1.1808424570008356,
+      "bytes": 2620310,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1527972080002655,
-      "bytes": 6634416,
-      "pages": 39,
+      "seconds": 1.1430494310006907,
+      "bytes": 10900083,
+      "pages": 37,
       "shards": 3
     },
     {
-      "seconds": 1.1540112700004101,
-      "bytes": 11807839,
-      "pages": 33,
+      "seconds": 1.591609420000168,
+      "bytes": 11822162,
+      "pages": 35,
       "shards": 3
     },
     {
-      "seconds": 0.8518780800004606,
-      "bytes": 2360936,
-      "pages": 30,
+      "seconds": 0.9329451240009803,
+      "bytes": 2485686,
+      "pages": 29,
       "shards": 2
     },
     {
-      "seconds": 1.2458887120010331,
-      "bytes": 6052831,
-      "pages": 35,
+      "seconds": 4.417064211000252,
+      "bytes": 6221614,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.0937011059995712,
-      "bytes": 11191807,
-      "pages": 40,
+      "seconds": 1.0521548180004174,
+      "bytes": 6267595,
+      "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.0167957749999914,
-      "bytes": 2251564,
+      "seconds": 0.9623302439977124,
+      "bytes": 6111937,
       "pages": 28,
       "shards": 2
     },
     {
-      "seconds": 0.9138518070012651,
-      "bytes": 6438645,
-      "pages": 35,
+      "seconds": 1.1251186529989354,
+      "bytes": 6276648,
+      "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 1.2295570830010547,
-      "bytes": 11846540,
-      "pages": 32,
+      "seconds": 1.272974397001235,
+      "bytes": 8064738,
+      "pages": 28,
       "shards": 3
     },
     {
-      "seconds": 0.9956274190008116,
-      "bytes": 6486340,
-      "pages": 41,
+      "seconds": 1.1465204630003427,
+      "bytes": 10636216,
+      "pages": 42,
       "shards": 3
     },
     {
-      "seconds": 1.1072240289995534,
-      "bytes": 6495721,
-      "pages": 37,
-      "shards": 2
+      "seconds": 0.9501123180016293,
+      "bytes": 1078327,
+      "pages": 34,
+      "shards": 1
     },
     {
-      "seconds": 0.8532074099985039,
-      "bytes": 2480365,
-      "pages": 29,
-      "shards": 2
+      "seconds": 3.8270818399978452,
+      "bytes": 7309721,
+      "pages": 35,
+      "shards": 4
     },
     {
-      "seconds": 0.959426125000391,
-      "bytes": 5985197,
-      "pages": 37,
+      "seconds": 1.1971988570003305,
+      "bytes": 10569382,
+      "pages": 35,
+      "shards": 3
+    },
+    {
+      "seconds": 0.8849704640015261,
+      "bytes": 2587633,
+      "pages": 35,
       "shards": 2
     },
     {
-      "seconds": 1.0452366619992972,
-      "bytes": 10510253,
+      "seconds": 2.1376686900002824,
+      "bytes": 11886724,
       "pages": 33,
       "shards": 3
     },
     {
-      "seconds": 1.1982196040007693,
-      "bytes": 8227634,
-      "pages": 32,
-      "shards": 3
+      "seconds": 1.1407382979996328,
+      "bytes": 5980776,
+      "pages": 35,
+      "shards": 2
     },
     {
-      "seconds": 1.881202160999237,
-      "bytes": 6307877,
-      "pages": 37,
+      "seconds": 0.9373651969981438,
+      "bytes": 2246840,
+      "pages": 25,
       "shards": 2
     },
     {
-      "seconds": 1.1178816689989617,
-      "bytes": 7521323,
+      "seconds": 2.0274646729994856,
+      "bytes": 10144494,
+      "pages": 44,
+      "shards": 3
+    },
+    {
+      "seconds": 2.5764055170002393,
+      "bytes": 7638759,
       "pages": 39,
       "shards": 3
     },
     {
-      "seconds": 1.323959568000646,
-      "bytes": 5092035,
+      "seconds": 0.8547688260005089,
+      "bytes": 858247,
+      "pages": 26,
+      "shards": 1
+    },
+    {
+      "seconds": 1.4054743929991673,
+      "bytes": 16596361,
       "pages": 32,
-      "shards": 3
+      "shards": 4
     },
     {
-      "seconds": 1.185996062000413,
-      "bytes": 6230664,
+      "seconds": 3.107019715000206,
+      "bytes": 2583987,
+      "pages": 34,
+      "shards": 2
+    },
+    {
+      "seconds": 1.155031209000299,
+      "bytes": 6268449,
       "pages": 36,
       "shards": 2
     },
     {
-      "seconds": 0.9238252899995132,
-      "bytes": 2424834,
-      "pages": 29,
+      "seconds": 0.8907613930023217,
+      "bytes": 6205142,
+      "pages": 37,
       "shards": 2
     },
     {
-      "seconds": 1.1364503580007295,
-      "bytes": 11698392,
+      "seconds": 1.1056236590011395,
+      "bytes": 6648792,
       "pages": 34,
       "shards": 3
-    },
-    {
-      "seconds": 1.0955314789989643,
-      "bytes": 11552086,
-      "pages": 37,
-      "shards": 4
     }
   ],
   "cache_key": {

trainer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ed5c494fe01b44806d348a5f83f4d51c818b578b1ae84966900fdecfbb1e84ce
-size 721756031
+oid sha256:f506a4eb49b59d5bfe154e01f4234e6f537368de7285d907d56b6e46d403a484
+size 721757439