appvoid commited on
Commit
02a2c27
·
verified ·
1 Parent(s): cd08a60

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": 76212.16586125136,
26
- "required_pages_per_second_upper_estimate": 0.0066058773788848285
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": 9526.52073265642,
40
- "required_pages_per_second_upper_estimate": 0.003800271554434506
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": 19053.04146531284,
54
- "required_pages_per_second_upper_estimate": 0.004407585820923007
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": 28579.562197969262,
68
- "required_pages_per_second_upper_estimate": 0.0017403453924818644
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": 28579.562197969262,
82
- "required_pages_per_second_upper_estimate": 0.0022001338112913136
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": 28579.562197969262,
96
- "required_pages_per_second_upper_estimate": 0.3548625128570628
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": 0.047263443851544906,
162
- "producer_p95_seconds": 0.14087222500029384,
163
- "gpu_update_seconds": 0.9674046022288104,
164
- "producer_batches_per_second": 21.158001163457598,
165
- "gpu_batches_per_second": 1.0336936558871983,
166
- "producer_headroom": 20.468347699491407,
167
  "sustainable": true,
168
- "buffer_seconds": 60.946489940415056,
169
  "buffer_drain_seconds": null,
170
  "shard_target_mib": 16,
171
- "upload_pages_per_commit": 71,
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": 75585.30080337779,
26
+ "required_pages_per_second_upper_estimate": 0.006551542304443311
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": 9448.162600422223,
40
+ "required_pages_per_second_upper_estimate": 0.003769013323927806
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": 18896.325200844447,
54
+ "required_pages_per_second_upper_estimate": 0.004371332271250298
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": 28344.48780126667,
68
+ "required_pages_per_second_upper_estimate": 0.0017260305950627186
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": 28344.48780126667,
82
+ "required_pages_per_second_upper_estimate": 0.00218203713350557
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": 28344.48780126667,
96
+ "required_pages_per_second_upper_estimate": 0.3519436755934126
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": 0.050977172562514994,
162
+ "producer_p95_seconds": 0.14861546000020098,
163
+ "gpu_update_seconds": 0.9754277513797394,
164
+ "producer_batches_per_second": 19.616623475412784,
165
+ "gpu_batches_per_second": 1.025191254386092,
166
+ "producer_headroom": 19.134598926284898,
167
  "sustainable": true,
168
+ "buffer_seconds": 60.47652058554384,
169
  "buffer_drain_seconds": null,
170
  "shard_target_mib": 16,
171
+ "upload_pages_per_commit": 72,
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
@@ -15,3 +15,4 @@
15
  {"step": 116500, "loss": 0.6138330940157175, "lr": 0.0003, "grad_norm": 0.13738930956550519, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17047979953, "tokens_per_second": 151677.26381402943, "data_wait_s": 1.4385999747901224e-05, "gpu_peak_gb": 85.46427917480469}
16
  {"step": 116600, "loss": 0.6223240718245506, "lr": 0.0003, "grad_norm": 0.14137559062344082, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17062709580, "tokens_per_second": 151335.11674431755, "data_wait_s": 1.1185000403202139e-05, "gpu_peak_gb": 85.46427917480469}
17
  {"step": 116700, "loss": 0.6237425673753023, "lr": 0.0003, "grad_norm": 0.13083552124961959, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17077478076, "tokens_per_second": 151831.34323416388, "data_wait_s": 1.7583000044396613e-05, "gpu_peak_gb": 85.46427917480469}
 
 
15
  {"step": 116500, "loss": 0.6138330940157175, "lr": 0.0003, "grad_norm": 0.13738930956550519, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17047979953, "tokens_per_second": 151677.26381402943, "data_wait_s": 1.4385999747901224e-05, "gpu_peak_gb": 85.46427917480469}
16
  {"step": 116600, "loss": 0.6223240718245506, "lr": 0.0003, "grad_norm": 0.14137559062344082, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17062709580, "tokens_per_second": 151335.11674431755, "data_wait_s": 1.1185000403202139e-05, "gpu_peak_gb": 85.46427917480469}
17
  {"step": 116700, "loss": 0.6237425673753023, "lr": 0.0003, "grad_norm": 0.13083552124961959, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17077478076, "tokens_per_second": 151831.34323416388, "data_wait_s": 1.7583000044396613e-05, "gpu_peak_gb": 85.46427917480469}
18
+ {"step": 116800, "loss": 0.6324888169765472, "lr": 0.0003, "grad_norm": 0.13479138816512448, "recovery_retries": 0, "loss_scale": 1.0, "microbatch": 86, "input_bytes_and_specials": 17092217704, "tokens_per_second": 151539.56402974177, "data_wait_s": 1.4374999864230631e-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 sha256:2c664544f2175702dcd2d00bb833d0dafd39452d6fbd3a99a00e0bf696a6e366
3
- size 36171
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7cf10f096b230700c5cf5f05eb277c3c2ea7356c5347da73f1f404beb58fe031
3
+ size 38289