| 12345678910 |
- c1-prose decode 30.50 tok/s (min 30.24 max 30.79 sd 0.18) ttft 0.297s toks [512, 512, 512, 512, 512] distinct 3 err 0 2026-09-24T03:33:36.560658Z..2026-09-24T03:35:01.787553Z
- c1-code decode 54.33 tok/s (min 54.16 max 54.62 sd 0.20) ttft 0.308s toks [512, 512, 512, 512, 512] distinct 1 err 0 2026-09-24T03:35:01.787910Z..2026-09-24T03:35:50.319550Z
- c1-chat_sampled decode 32.98 tok/s (min 32.26 max 34.61 sd 0.84) ttft 0.273s toks [512, 512, 512, 512, 512] distinct 5 err 0 2026-09-24T03:35:50.319819Z..2026-09-24T03:37:08.454057Z
- c1-prose2 decode 27.93 tok/s (min 27.91 max 28.33 sd 0.16) ttft 0.280s toks [512, 512, 512, 512, 512] distinct 2 err 0 2026-09-24T03:37:08.454348Z..2026-09-24T03:38:41.014076Z
- c4 agg 69.47 tok/s (min 68.33 max 69.57) per-stream 23.29 2026-09-24T03:38:41.014255Z..2026-09-24T03:41:09.217891Z
- c1-prose accept_len 2.10 step 68.3 ms (n=31 intervals, bs=1)
- c1-code accept_len 4.30 step 77.5 ms (n=16 intervals, bs=1.0)
- c1-chat_sampled accept_len 2.34 step 69.1 ms (n=28 intervals, bs=1.0)
- c1-prose2 accept_len 1.82 step 65.0 ms (n=35 intervals, bs=1)
- c4 accept_len 2.37 step 113.3 ms (n=32 intervals, bs=3.0)
|
