appvoid commited on
Commit
164af9e
·
verified ·
1 Parent(s): a69fe3e

Qwen90M step 5000; complete resumable checkpoint

Browse files
Files changed (7) hide show
  1. complete.json +6 -6
  2. coverage.json +80 -27
  3. latest.json +1 -1
  4. model.safetensors +1 -1
  5. shard_scheduler.json +287 -29
  6. trainer.pt +2 -2
  7. training_config.json +2 -2
complete.json CHANGED
@@ -1,18 +1,18 @@
1
  {
2
- "step": 0,
3
  "files": {
4
  "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
5
  "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
6
- "model.safetensors": "1c688d2ef10bd45dcb2ed3ca6945c4cc383ba4cc356d90bbb93b725e427338c1",
7
  "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
8
  "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
9
  "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
10
  "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
11
- "trainer.pt": "21aa2cccd888f3918a44b02c1ad179a74c9924c4304886988affaf97c4132422",
12
- "training_config.json": "528161bcd9687da465595131ceb9bfb223ac3e494a1c6d5fcfdbd97b88324bd8",
13
  "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e",
14
- "shard_scheduler.json": "c16912023ae706fc358e0de73ec412ec549d6d440abc128e7900e41c5c63a6f3",
15
- "coverage.json": "1db676f88d72dde495669f8f42a2a0104a6dc4ec5bfa810967263316d2335d79",
16
  "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
17
  },
18
  "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"
 
1
  {
2
+ "step": 5000,
3
  "files": {
4
  "config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7",
5
  "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4",
6
+ "model.safetensors": "c8aa1507ab6f726c5ffd1b20a36284bd30c32cd1cf5c419d73cbc66d98ed6afd",
7
  "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03",
8
  "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300",
9
  "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263",
10
  "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6",
11
+ "trainer.pt": "bc1fa5664111a090e684ef00bc5753f106bfd58076a5047bfda30955ab3d9b43",
12
+ "training_config.json": "f08d2b46b36e5a060b47b42699c2a03ce255f88d1589e7e6a18a4b35ab59c4c7",
13
  "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e",
14
+ "shard_scheduler.json": "fe416c1fb82665d94c94be2f3fbdb92516402dd5bee1d4df92808e6efd9f97ab",
15
+ "coverage.json": "6a692152610fe0c0340ddcdb7db228594b1270a850305aaeacbb81308e2c0c68",
16
  "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"
17
  },
18
  "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063"
coverage.json CHANGED
@@ -1,54 +1,107 @@
1
  {
2
- "step": 0,
3
- "stats": {},
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
  "sources": {
5
  "climbmix": {
6
- "page": 0,
7
- "row": 0,
8
- "start": null,
 
 
 
 
 
 
 
9
  "epoch": 0,
10
- "offset": 0,
11
- "documents": 0
12
  },
13
  "rewrite6": {
14
- "page": 0,
15
- "row": 0,
16
- "start": null,
 
 
 
 
 
 
17
  "epoch": 0,
18
  "offset": 0,
19
- "documents": 0
20
  },
21
  "openmath": {
22
- "page": 0,
23
- "row": 0,
24
- "start": null,
 
 
 
 
 
 
 
25
  "epoch": 0,
26
  "offset": 0,
27
- "documents": 0
28
  },
29
  "ultra_qa": {
30
- "page": 0,
31
- "row": 0,
32
- "start": null,
 
 
 
 
 
 
 
33
  "epoch": 0,
34
  "offset": 0,
35
- "documents": 0
36
  },
37
  "ultra_style": {
38
- "page": 0,
39
- "row": 0,
40
- "start": null,
 
 
 
 
 
 
 
41
  "epoch": 0,
42
- "offset": 0,
43
- "documents": 0
44
  },
45
  "cortex": {
46
- "page": 0,
47
- "row": 0,
48
  "start": null,
49
  "epoch": 0,
50
  "offset": 0,
51
- "documents": 0
52
  }
53
  }
54
  }
 
1
  {
2
+ "step": 5000,
3
+ "stats": {
4
+ "climbmix/blocks": 180000,
5
+ "climbmix/targets": 300287575,
6
+ "climbmix/input_tokens": 300306959,
7
+ "ultra_qa/blocks": 67500,
8
+ "ultra_qa/targets": 110133457,
9
+ "ultra_qa/input_tokens": 110133662,
10
+ "cortex/blocks": 67500,
11
+ "cortex/targets": 2571669,
12
+ "cortex/input_tokens": 107334831,
13
+ "ultra_style/blocks": 67500,
14
+ "ultra_style/targets": 106887961,
15
+ "ultra_style/input_tokens": 106893821,
16
+ "openmath/blocks": 45000,
17
+ "openmath/targets": 69552253,
18
+ "openmath/input_tokens": 69575649,
19
+ "rewrite6/blocks": 22500,
20
+ "rewrite6/targets": 36855636,
21
+ "rewrite6/input_tokens": 36896125
22
+ },
23
  "sources": {
24
  "climbmix": {
25
+ "page": 24,
26
+ "row": 3774,
27
+ "start": {
28
+ "file": 1,
29
+ "reader": {
30
+ "format": "parquet-row-v1",
31
+ "row_group": 11,
32
+ "row": 1024
33
+ }
34
+ },
35
  "epoch": 0,
36
+ "offset": 2048,
37
+ "documents": 102078
38
  },
39
  "rewrite6": {
40
+ "page": 14,
41
+ "row": 277,
42
+ "start": {
43
+ "file": 0,
44
+ "reader": {
45
+ "format": "jsonl-row-v1",
46
+ "row": 57344
47
+ }
48
+ },
49
  "epoch": 0,
50
  "offset": 0,
51
+ "documents": 57621
52
  },
53
  "openmath": {
54
+ "page": 15,
55
+ "row": 1897,
56
+ "start": {
57
+ "file": 0,
58
+ "reader": {
59
+ "format": "parquet-row-v1",
60
+ "row_group": 61,
61
+ "row": 440
62
+ }
63
+ },
64
  "epoch": 0,
65
  "offset": 0,
66
+ "documents": 63337
67
  },
68
  "ultra_qa": {
69
+ "page": 6,
70
+ "row": 2707,
71
+ "start": {
72
+ "file": 0,
73
+ "reader": {
74
+ "format": "parquet-row-v1",
75
+ "row_group": 0,
76
+ "row": 24576
77
+ }
78
+ },
79
  "epoch": 0,
80
  "offset": 0,
81
+ "documents": 27283
82
  },
83
  "ultra_style": {
84
+ "page": 10,
85
+ "row": 1000,
86
+ "start": {
87
+ "file": 0,
88
+ "reader": {
89
+ "format": "parquet-row-v1",
90
+ "row_group": 0,
91
+ "row": 40960
92
+ }
93
+ },
94
  "epoch": 0,
95
+ "offset": 2048,
96
+ "documents": 41960
97
  },
98
  "cortex": {
99
+ "page": 1309,
100
+ "row": 72,
101
  "start": null,
102
  "epoch": 0,
103
  "offset": 0,
104
+ "documents": 167624
105
  }
106
  }
107
  }
latest.json CHANGED
@@ -1 +1 @@
1
- {"step": 0, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "1c688d2ef10bd45dcb2ed3ca6945c4cc383ba4cc356d90bbb93b725e427338c1", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "21aa2cccd888f3918a44b02c1ad179a74c9924c4304886988affaf97c4132422", "training_config.json": "528161bcd9687da465595131ceb9bfb223ac3e494a1c6d5fcfdbd97b88324bd8", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "c16912023ae706fc358e0de73ec412ec549d6d440abc128e7900e41c5c63a6f3", "coverage.json": "1db676f88d72dde495669f8f42a2a0104a6dc4ec5bfa810967263316d2335d79", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000000000"}
 
1
+ {"step": 5000, "files": {"config.json": "cbc72137da3dae65085a2a8c39ad7634d16bf92d785ca00d283ee82fe32b37e7", "generation_config.json": "54ed0d19f6226888c91157964ede8a892bd5ae175bfad85336152b3d9b6cd7c4", "model.safetensors": "c8aa1507ab6f726c5ffd1b20a36284bd30c32cd1cf5c419d73cbc66d98ed6afd", "tokenization_byte.py": "279b9990f77b1eaedc7c1184fe13d60273a713ead856200f1fb9ae323c7b1e03", "tokenizer_config.json": "0ecbd4fd39e35125800d806af981ba0cacd8a7ba5f02c9e729c4ecc54af6d300", "special_tokens_map.json": "20d4b94d0e657c3f44d9123522c106e7315b6c1a1b6d259478f17bd2918b1263", "byte_vocab.json": "57a1cd2e70dd5b7cc6fd2eaa12ebfcd173c1620fa5d45c2c6282511c29b6a5f6", "trainer.pt": "bc1fa5664111a090e684ef00bc5753f106bfd58076a5047bfda30955ab3d9b43", "training_config.json": "f08d2b46b36e5a060b47b42699c2a03ce255f88d1589e7e6a18a4b35ab59c4c7", "performance.json": "b95a1befd575f8d9a02b43fd0cc41c54243f5297127b93d2656e0292ba15e77e", "shard_scheduler.json": "fe416c1fb82665d94c94be2f3fbdb92516402dd5bee1d4df92808e6efd9f97ab", "coverage.json": "6a692152610fe0c0340ddcdb7db228594b1270a850305aaeacbb81308e2c0c68", "README.md": "9ce0f9efacbc46177a7e818b6aed274d84e7145ff944b5c7d6ecd5d33b9d4419"}, "signature": "9fee47918905b0750bee87876aa424f021ccc4507ee8db22a11d01fbc6109063", "folder": "step-000000005000"}
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1c688d2ef10bd45dcb2ed3ca6945c4cc383ba4cc356d90bbb93b725e427338c1
3
  size 360622744
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c8aa1507ab6f726c5ffd1b20a36284bd30c32cd1cf5c419d73cbc66d98ed6afd
3
  size 360622744
shard_scheduler.json CHANGED
@@ -22,8 +22,8 @@
22
  "tokens_per_second": 1062967.6168812835,
23
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
24
  "estimated_pages_for_buffer": 1,
25
- "required_token_positions_per_second": 78853.96613762408,
26
- "required_pages_per_second_upper_estimate": 0.00683486193125923
27
  },
28
  {
29
  "source": "rewrite6",
@@ -36,8 +36,8 @@
36
  "tokens_per_second": 1206672.263377472,
37
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
38
  "estimated_pages_for_buffer": 1,
39
- "required_token_positions_per_second": 9856.74576720301,
40
- "required_pages_per_second_upper_estimate": 0.003932003258019391
41
  },
42
  {
43
  "source": "openmath",
@@ -50,8 +50,8 @@
50
  "tokens_per_second": 410042.7581134458,
51
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
52
  "estimated_pages_for_buffer": 1,
53
- "required_token_positions_per_second": 19713.49153440602,
54
- "required_pages_per_second_upper_estimate": 0.004560369320883491
55
  },
56
  {
57
  "source": "ultra_qa",
@@ -64,8 +64,8 @@
64
  "tokens_per_second": 2163486.5624068812,
65
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
66
  "estimated_pages_for_buffer": 1,
67
- "required_token_positions_per_second": 29570.23730160903,
68
- "required_pages_per_second_upper_estimate": 0.0018006723086229548
69
  },
70
  {
71
  "source": "ultra_style",
@@ -78,8 +78,8 @@
78
  "tokens_per_second": 1546694.2737633225,
79
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
80
  "estimated_pages_for_buffer": 1,
81
- "required_token_positions_per_second": 29570.23730160903,
82
- "required_pages_per_second_upper_estimate": 0.0022763987231337094
83
  },
84
  {
85
  "source": "cortex",
@@ -91,9 +91,9 @@
91
  "read_seconds": 0.042780299000014566,
92
  "tokens_per_second": 1015177.8395898147,
93
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
94
- "estimated_pages_for_buffer": 23,
95
- "required_token_positions_per_second": 29570.23730160903,
96
- "required_pages_per_second_upper_estimate": 0.36716338206798155
97
  }
98
  ],
99
  "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -155,27 +155,27 @@
155
  "shard_benchmark_scope": "Local packing/read measurements; upload capacity measured only on real data commits. Targets larger than available pages are sample-limited.",
156
  "estimated_fresh_generation_batch_seconds": 0.1434574083060457,
157
  "selected": {
158
- "prefetch_batches": 65,
159
  "prefetch_capacity_limit": 512,
160
  "batch_memory_estimate_bytes": 688946,
161
- "producer_batch_seconds": 2.3582960475177965,
162
- "producer_p95_seconds": 17.862166522000052,
163
- "gpu_update_seconds": 0.9349941874999956,
164
- "producer_batches_per_second": 0.4240349726458394,
165
- "gpu_batches_per_second": 1.06952536536491,
166
- "producer_headroom": 0.39647023472057946,
167
- "sustainable": false,
168
- "buffer_seconds": 60.774622187499716,
169
- "buffer_drain_seconds": 100.69863274989008,
170
- "shard_target_mib": 32,
171
- "upload_pages_per_commit": 15,
172
- "max_pending_pages": 1842,
173
- "generation_bytes_per_second": 1035119.2626246791,
174
- "upload_bytes_per_second": 6146691.738378885,
175
  "upload_sustainable": true,
176
  "logical_page_rows": 4096,
177
  "logical_cortex_page_rows": 128,
178
- "estimated_full_shards_per_hour": 111.0562487080349,
179
  "selection_policy": "Smallest shard within 90% of measured local peak, enlarged to amortize 30 seconds of measured page production; bounded to 256 MiB."
180
  },
181
  "upload_measurements": [
@@ -184,6 +184,264 @@
184
  "bytes": 5047481,
185
  "pages": 1,
186
  "shards": 1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
187
  }
188
  ],
189
  "cache_key": {
 
22
  "tokens_per_second": 1062967.6168812835,
23
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
24
  "estimated_pages_for_buffer": 1,
25
+ "required_token_positions_per_second": 74279.87317241088,
26
+ "required_pages_per_second_upper_estimate": 0.006438391145967182
27
  },
28
  {
29
  "source": "rewrite6",
 
36
  "tokens_per_second": 1206672.263377472,
37
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
38
  "estimated_pages_for_buffer": 1,
39
+ "required_token_positions_per_second": 9284.98414655136,
40
+ "required_pages_per_second_upper_estimate": 0.003703918998943418
41
  },
42
  {
43
  "source": "openmath",
 
50
  "tokens_per_second": 410042.7581134458,
51
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
52
  "estimated_pages_for_buffer": 1,
53
+ "required_token_positions_per_second": 18569.96829310272,
54
+ "required_pages_per_second_upper_estimate": 0.004295835344329655
55
  },
56
  {
57
  "source": "ultra_qa",
 
64
  "tokens_per_second": 2163486.5624068812,
65
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
66
  "estimated_pages_for_buffer": 1,
67
+ "required_token_positions_per_second": 27854.95243965408,
68
+ "required_pages_per_second_upper_estimate": 0.0016962204599340583
69
  },
70
  {
71
  "source": "ultra_style",
 
78
  "tokens_per_second": 1546694.2737633225,
79
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
80
  "estimated_pages_for_buffer": 1,
81
+ "required_token_positions_per_second": 27854.95243965408,
82
+ "required_pages_per_second_upper_estimate": 0.00214435134624802
83
  },
84
  {
85
  "source": "cortex",
 
91
  "read_seconds": 0.042780299000014566,
92
  "tokens_per_second": 1015177.8395898147,
93
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
94
+ "estimated_pages_for_buffer": 21,
95
+ "required_token_positions_per_second": 27854.95243965408,
96
+ "required_pages_per_second_upper_estimate": 0.3458652847716463
97
  }
98
  ],
99
  "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
 
155
  "shard_benchmark_scope": "Local packing/read measurements; upload capacity measured only on real data commits. Targets larger than available pages are sample-limited.",
156
  "estimated_fresh_generation_batch_seconds": 0.1434574083060457,
157
  "selected": {
158
+ "prefetch_batches": 61,
159
  "prefetch_capacity_limit": 512,
160
  "batch_memory_estimate_bytes": 688946,
161
+ "producer_batch_seconds": 0.2317790046171737,
162
+ "producer_p95_seconds": 0.17695051500049885,
163
+ "gpu_update_seconds": 0.992570353867865,
164
+ "producer_batches_per_second": 4.314454631694042,
165
+ "gpu_batches_per_second": 1.007485258957396,
166
+ "producer_headroom": 4.2823997605274045,
167
+ "sustainable": true,
168
+ "buffer_seconds": 60.546791585939765,
169
+ "buffer_drain_seconds": null,
170
+ "shard_target_mib": 16,
171
+ "upload_pages_per_commit": 79,
172
+ "max_pending_pages": 4096,
173
+ "generation_bytes_per_second": 328113.0883383849,
174
+ "upload_bytes_per_second": 4268388.328580749,
175
  "upload_sustainable": true,
176
  "logical_page_rows": 4096,
177
  "logical_cortex_page_rows": 128,
178
+ "estimated_full_shards_per_hour": 70.40543067563688,
179
  "selection_policy": "Smallest shard within 90% of measured local peak, enlarged to amortize 30 seconds of measured page production; bounded to 256 MiB."
180
  },
181
  "upload_measurements": [
 
184
  "bytes": 5047481,
185
  "pages": 1,
186
  "shards": 1
187
+ },
188
+ {
189
+ "seconds": 2.1621551530000716,
190
+ "bytes": 13914829,
191
+ "pages": 15,
192
+ "shards": 5
193
+ },
194
+ {
195
+ "seconds": 0.7754683500000965,
196
+ "bytes": 204111,
197
+ "pages": 6,
198
+ "shards": 1
199
+ },
200
+ {
201
+ "seconds": 1.1240084219999744,
202
+ "bytes": 42370,
203
+ "pages": 1,
204
+ "shards": 1
205
+ },
206
+ {
207
+ "seconds": 1.1000538829999869,
208
+ "bytes": 494707,
209
+ "pages": 15,
210
+ "shards": 1
211
+ },
212
+ {
213
+ "seconds": 0.9781999550000364,
214
+ "bytes": 5863999,
215
+ "pages": 27,
216
+ "shards": 2
217
+ },
218
+ {
219
+ "seconds": 0.9417934939999668,
220
+ "bytes": 2584042,
221
+ "pages": 34,
222
+ "shards": 2
223
+ },
224
+ {
225
+ "seconds": 0.966999364000003,
226
+ "bytes": 6885855,
227
+ "pages": 36,
228
+ "shards": 3
229
+ },
230
+ {
231
+ "seconds": 1.0705818949998047,
232
+ "bytes": 10769730,
233
+ "pages": 37,
234
+ "shards": 3
235
+ },
236
+ {
237
+ "seconds": 1.2535849739999776,
238
+ "bytes": 3640398,
239
+ "pages": 33,
240
+ "shards": 3
241
+ },
242
+ {
243
+ "seconds": 1.3395652879999034,
244
+ "bytes": 11637618,
245
+ "pages": 33,
246
+ "shards": 3
247
+ },
248
+ {
249
+ "seconds": 0.9964154829999643,
250
+ "bytes": 2622281,
251
+ "pages": 35,
252
+ "shards": 2
253
+ },
254
+ {
255
+ "seconds": 1.0029198830000041,
256
+ "bytes": 11222438,
257
+ "pages": 33,
258
+ "shards": 4
259
+ },
260
+ {
261
+ "seconds": 1.2415246609998576,
262
+ "bytes": 5982796,
263
+ "pages": 34,
264
+ "shards": 2
265
+ },
266
+ {
267
+ "seconds": 1.1530046430000311,
268
+ "bytes": 2613544,
269
+ "pages": 34,
270
+ "shards": 2
271
+ },
272
+ {
273
+ "seconds": 22.672299763999945,
274
+ "bytes": 7363447,
275
+ "pages": 38,
276
+ "shards": 3
277
+ },
278
+ {
279
+ "seconds": 1.0496788179998475,
280
+ "bytes": 15965043,
281
+ "pages": 34,
282
+ "shards": 4
283
+ },
284
+ {
285
+ "seconds": 0.8725824939997437,
286
+ "bytes": 2480897,
287
+ "pages": 35,
288
+ "shards": 2
289
+ },
290
+ {
291
+ "seconds": 1.156973391000065,
292
+ "bytes": 7369438,
293
+ "pages": 38,
294
+ "shards": 3
295
+ },
296
+ {
297
+ "seconds": 0.9484031940000932,
298
+ "bytes": 2544896,
299
+ "pages": 35,
300
+ "shards": 2
301
+ },
302
+ {
303
+ "seconds": 1.0683630309999899,
304
+ "bytes": 9860015,
305
+ "pages": 31,
306
+ "shards": 3
307
+ },
308
+ {
309
+ "seconds": 0.8969114139999874,
310
+ "bytes": 7503017,
311
+ "pages": 38,
312
+ "shards": 3
313
+ },
314
+ {
315
+ "seconds": 0.8951566169998841,
316
+ "bytes": 8341517,
317
+ "pages": 32,
318
+ "shards": 3
319
+ },
320
+ {
321
+ "seconds": 1.104310605000137,
322
+ "bytes": 5895784,
323
+ "pages": 34,
324
+ "shards": 2
325
+ },
326
+ {
327
+ "seconds": 1.2573788659997263,
328
+ "bytes": 8080667,
329
+ "pages": 37,
330
+ "shards": 4
331
+ },
332
+ {
333
+ "seconds": 0.9144526599998244,
334
+ "bytes": 5532432,
335
+ "pages": 26,
336
+ "shards": 2
337
+ },
338
+ {
339
+ "seconds": 0.9253262060001362,
340
+ "bytes": 5975019,
341
+ "pages": 35,
342
+ "shards": 2
343
+ },
344
+ {
345
+ "seconds": 0.8811729529998047,
346
+ "bytes": 3757694,
347
+ "pages": 36,
348
+ "shards": 3
349
+ },
350
+ {
351
+ "seconds": 1.5062500680000994,
352
+ "bytes": 15040116,
353
+ "pages": 34,
354
+ "shards": 4
355
+ },
356
+ {
357
+ "seconds": 1.0638150449999557,
358
+ "bytes": 7164962,
359
+ "pages": 39,
360
+ "shards": 3
361
+ },
362
+ {
363
+ "seconds": 1.0547767480002221,
364
+ "bytes": 2359076,
365
+ "pages": 27,
366
+ "shards": 2
367
+ },
368
+ {
369
+ "seconds": 1.027329577999808,
370
+ "bytes": 11020380,
371
+ "pages": 34,
372
+ "shards": 3
373
+ },
374
+ {
375
+ "seconds": 1.2178119280006285,
376
+ "bytes": 3672261,
377
+ "pages": 37,
378
+ "shards": 3
379
+ },
380
+ {
381
+ "seconds": 1.0616448490000039,
382
+ "bytes": 5911843,
383
+ "pages": 30,
384
+ "shards": 2
385
+ },
386
+ {
387
+ "seconds": 1.2120231849994525,
388
+ "bytes": 11658602,
389
+ "pages": 33,
390
+ "shards": 3
391
+ },
392
+ {
393
+ "seconds": 0.9783495700003186,
394
+ "bytes": 7342912,
395
+ "pages": 40,
396
+ "shards": 4
397
+ },
398
+ {
399
+ "seconds": 0.9914971539992621,
400
+ "bytes": 6238955,
401
+ "pages": 35,
402
+ "shards": 2
403
+ },
404
+ {
405
+ "seconds": 0.9600033409997195,
406
+ "bytes": 1092240,
407
+ "pages": 31,
408
+ "shards": 1
409
+ },
410
+ {
411
+ "seconds": 0.9772284549999313,
412
+ "bytes": 8526326,
413
+ "pages": 38,
414
+ "shards": 4
415
+ },
416
+ {
417
+ "seconds": 1.260473302999344,
418
+ "bytes": 10994776,
419
+ "pages": 35,
420
+ "shards": 3
421
+ },
422
+ {
423
+ "seconds": 0.9115331740003967,
424
+ "bytes": 8005286,
425
+ "pages": 27,
426
+ "shards": 3
427
+ },
428
+ {
429
+ "seconds": 0.9674303089996101,
430
+ "bytes": 7444249,
431
+ "pages": 41,
432
+ "shards": 3
433
+ },
434
+ {
435
+ "seconds": 0.8210835840000072,
436
+ "bytes": 900296,
437
+ "pages": 25,
438
+ "shards": 1
439
+ },
440
+ {
441
+ "seconds": 1.093376490999617,
442
+ "bytes": 11563576,
443
+ "pages": 39,
444
+ "shards": 4
445
  }
446
  ],
447
  "cache_key": {
trainer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:21aa2cccd888f3918a44b02c1ad179a74c9924c4304886988affaf97c4132422
3
- size 340197
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bc1fa5664111a090e684ef00bc5753f106bfd58076a5047bfda30955ab3d9b43
3
+ size 721764031
training_config.json CHANGED
@@ -46,8 +46,8 @@
46
  "cpu_threads": 4,
47
  "page_rows": 4096,
48
  "cortex_page_rows": 128,
49
- "prefetch_batches": 65,
50
- "prefetch_low_water": 32,
51
  "memory_pages": 12,
52
  "cache_gb": 20,
53
  "force_rebenchmark": false,
 
46
  "cpu_threads": 4,
47
  "page_rows": 4096,
48
  "cortex_page_rows": 128,
49
+ "prefetch_batches": 61,
50
+ "prefetch_low_water": 30,
51
  "memory_pages": 12,
52
  "cache_gb": 20,
53
  "force_rebenchmark": false,