appvoid commited on
Commit
ba06a4d
·
verified ·
1 Parent(s): b1142f5

Update TensorBoard training metrics

Browse files
benchmark_cache/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": 76263.54012313594,
26
- "required_pages_per_second_upper_estimate": 0.006610330369698108
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": 9532.942515391993,
40
- "required_pages_per_second_upper_estimate": 0.003802833299581934
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": 19065.885030783986,
54
- "required_pages_per_second_upper_estimate": 0.004410556953755724
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": 28598.827546175977,
68
- "required_pages_per_second_upper_estimate": 0.001741518551110183
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": 28598.827546175977,
82
- "required_pages_per_second_upper_estimate": 0.0022016169111261637
83
  },
84
  {
85
  "source": "cortex",
@@ -92,8 +92,8 @@
92
  "tokens_per_second": 1015177.8395898147,
93
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
94
  "estimated_pages_for_buffer": 22,
95
- "required_token_positions_per_second": 28598.827546175977,
96
- "required_pages_per_second_upper_estimate": 0.3551017240048174
97
  }
98
  ],
99
  "gpu_measurement": "Selected forward/backward estimate; replaced by live successful update timings.",
@@ -155,20 +155,20 @@
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": 63,
159
  "prefetch_capacity_limit": 512,
160
  "batch_memory_estimate_bytes": 688946,
161
- "producer_batch_seconds": 0.04772247081250569,
162
- "producer_p95_seconds": 0.13767005699992296,
163
- "gpu_update_seconds": 0.9667529186418303,
164
- "producer_batches_per_second": 20.95448921596804,
165
- "gpu_batches_per_second": 1.0343904639097214,
166
- "producer_headroom": 20.25781360818586,
167
  "sustainable": true,
168
- "buffer_seconds": 60.90543387443531,
169
  "buffer_drain_seconds": null,
170
  "shard_target_mib": 16,
171
- "upload_pages_per_commit": 80,
172
  "max_pending_pages": 4096,
173
  "generation_bytes_per_second": 0,
174
  "upload_bytes_per_second": null,
 
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": 76166.02571118445,
26
+ "required_pages_per_second_upper_estimate": 0.00660187806761817
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": 9520.753213898057,
40
+ "required_pages_per_second_upper_estimate": 0.003797970804969705
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": 19041.506427796114,
54
+ "required_pages_per_second_upper_estimate": 0.004404917393003239
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": 28562.25964169417,
68
+ "required_pages_per_second_upper_estimate": 0.0017392917575842083
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": 28562.25964169417,
82
+ "required_pages_per_second_upper_estimate": 0.002198801812612716
83
  },
84
  {
85
  "source": "cortex",
 
92
  "tokens_per_second": 1015177.8395898147,
93
  "note": "Bounded fresh sample; full-page latency is an extrapolation.",
94
  "estimated_pages_for_buffer": 22,
95
+ "required_token_positions_per_second": 28562.25964169417,
96
+ "required_pages_per_second_upper_estimate": 0.3546476730160569
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": 62,
159
  "prefetch_capacity_limit": 512,
160
  "batch_memory_estimate_bytes": 688946,
161
+ "producer_batch_seconds": 0.05994781818750994,
162
+ "producer_p95_seconds": 0.15240313300000707,
163
+ "gpu_update_seconds": 0.9679906403357679,
164
+ "producer_batches_per_second": 16.6811742317646,
165
+ "gpu_batches_per_second": 1.0330678400497024,
166
+ "producer_headroom": 16.147220526158325,
167
  "sustainable": true,
168
+ "buffer_seconds": 60.01541970081761,
169
  "buffer_drain_seconds": null,
170
  "shard_target_mib": 16,
171
+ "upload_pages_per_commit": 81,
172
  "max_pending_pages": 4096,
173
  "generation_bytes_per_second": 0,
174
  "upload_bytes_per_second": null,
metrics/void-byte-v1/1790519810984249790-cb9c9513/train.jsonl CHANGED
@@ -26,3 +26,4 @@
26
  {"step": 117600, "loss": 0.6016160920262337, "lr": 0.0003, "grad_norm": 0.1248867513326305, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17209104158, "tokens_per_second": 151913.17321309296, "data_wait_s": 1.3078000392852118e-05, "gpu_peak_gb": 85.46427917480469}
27
  {"step": 117700, "loss": 0.6191273359581828, "lr": 0.0003, "grad_norm": 0.16089004303991808, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17223795313, "tokens_per_second": 151289.57461195142, "data_wait_s": 1.3388000297709368e-05, "gpu_peak_gb": 85.46427917480469}
28
  {"step": 117800, "loss": 0.5868233777582645, "lr": 0.0003, "grad_norm": 0.13449859552857432, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17238468392, "tokens_per_second": 150951.86868817345, "data_wait_s": 1.8103000002156477e-05, "gpu_peak_gb": 85.46427917480469}
 
 
26
  {"step": 117600, "loss": 0.6016160920262337, "lr": 0.0003, "grad_norm": 0.1248867513326305, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17209104158, "tokens_per_second": 151913.17321309296, "data_wait_s": 1.3078000392852118e-05, "gpu_peak_gb": 85.46427917480469}
27
  {"step": 117700, "loss": 0.6191273359581828, "lr": 0.0003, "grad_norm": 0.16089004303991808, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17223795313, "tokens_per_second": 151289.57461195142, "data_wait_s": 1.3388000297709368e-05, "gpu_peak_gb": 85.46427917480469}
28
  {"step": 117800, "loss": 0.5868233777582645, "lr": 0.0003, "grad_norm": 0.13449859552857432, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17238468392, "tokens_per_second": 150951.86868817345, "data_wait_s": 1.8103000002156477e-05, "gpu_peak_gb": 85.46427917480469}
29
+ {"step": 117900, "loss": 0.6241789255291224, "lr": 0.0003, "grad_norm": 0.1445657070540139, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17253039451, "tokens_per_second": 149567.94261848138, "data_wait_s": 1.4555999769072514e-05, "gpu_peak_gb": 85.46427917480469}
runs/void-byte-v1/events.out.tfevents.1790519810.cad0b19a-59f2-43f0-8fb4-2a94097843e2-tjh7c.1336.0.1790519810984249790-cb9c9513 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fc55604c411802102431d340b451f0a5a2199e06bcb7ca301ba36c5e25f26310
3
- size 59469
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:78feca4ad1b945bb7c09d3be439f09696ec1f043a5bf40ce27056edbb8594d81
3
+ size 61587