1.8 KiB
1.8 KiB
| 1 | ctx_tokens | prefill_tokens | prefill_tps | gen_tokens | gen_tps | gen_first_ms | gen_steady_tokens | gen_steady_tps | kvcache_bytes |
|---|---|---|---|---|---|---|---|---|---|
| 2 | 2048 | 2048 | 790.18 | 128 | 39.35 | 26.947 | 127 | 40.00 | 52184460 |
| 3 | 4096 | 2048 | 710.50 | 128 | 37.50 | 28.069 | 127 | 38.09 | 80373132 |
| 4 | 6144 | 2048 | 704.17 | 128 | 37.18 | 30.710 | 127 | 37.79 | 108561804 |
| 5 | 8192 | 2048 | 683.85 | 128 | 37.03 | 30.095 | 127 | 37.63 | 136750476 |
| 6 | 10240 | 2048 | 651.30 | 128 | 36.76 | 30.129 | 127 | 37.35 | 164939148 |
| 7 | 12288 | 2048 | 625.68 | 128 | 36.63 | 30.317 | 127 | 37.21 | 193127820 |
| 8 | 14336 | 2048 | 588.80 | 128 | 36.40 | 28.378 | 127 | 36.95 | 221316492 |
| 9 | 16384 | 2048 | 572.53 | 128 | 36.14 | 28.536 | 127 | 36.68 | 249505164 |
| 10 | 18432 | 2048 | 565.67 | 128 | 35.79 | 28.967 | 127 | 36.32 | 277693836 |
| 11 | 20480 | 2048 | 566.02 | 128 | 35.71 | 29.048 | 127 | 36.24 | 305882508 |
| 12 | 22528 | 2048 | 567.67 | 128 | 35.43 | 28.986 | 127 | 35.94 | 334071180 |
| 13 | 24576 | 2048 | 570.39 | 128 | 35.35 | 30.440 | 127 | 35.88 | 362259852 |
| 14 | 26624 | 2048 | 567.38 | 128 | 35.02 | 29.618 | 127 | 35.52 | 390448524 |
| 15 | 28672 | 2048 | 569.86 | 128 | 34.90 | 29.328 | 127 | 35.40 | 418637196 |
| 16 | 30720 | 2048 | 570.69 | 128 | 34.68 | 30.429 | 127 | 35.19 | 446825868 |
| 17 | 32768 | 2048 | 557.04 | 128 | 34.36 | 35.688 | 127 | 34.90 | 475014540 |
| 18 | 34816 | 2048 | 534.63 | 128 | 34.00 | 37.818 | 127 | 34.55 | 503203212 |
| 19 | 36864 | 2048 | 510.85 | 128 | 33.71 | 38.606 | 127 | 34.25 | 531391884 |
| 20 | 38912 | 2048 | 498.35 | 128 | 32.71 | 35.003 | 127 | 33.18 | 559580556 |
| 21 | 40960 | 2048 | 489.46 | 128 | 32.05 | 35.391 | 127 | 32.50 | 587769228 |
| 22 | 43008 | 2048 | 475.25 | 128 | 31.85 | 38.767 | 127 | 32.32 | 615957900 |
| 23 | 45056 | 2048 | 465.50 | 128 | 31.31 | 39.407 | 127 | 31.76 | 644146572 |
| 24 | 47104 | 2048 | 455.29 | 128 | 30.46 | 43.102 | 127 | 30.92 | 672335244 |
| 25 | 49152 | 2048 | 448.53 | 128 | 29.07 | 45.980 | 127 | 29.50 | 700523916 |
| 26 | 51200 | 2048 | 435.31 | 128 | 28.75 | 46.071 | 127 | 29.16 | 728712588 |
| 27 | 53248 | 2048 | 430.93 | 128 | 28.36 | 43.866 | 127 | 28.74 | 756901260 |
| 28 | 55296 | 2048 | 420.30 | 128 | 28.34 | 44.054 | 127 | 28.73 | 785089932 |
| 29 | 57344 | 2048 | 420.71 | 128 | 27.87 | 44.341 | 127 | 28.24 | 813278604 |
| 30 | 59392 | 2048 | 407.01 | 128 | 27.96 | 44.062 | 127 | 28.33 | 841467276 |
| 31 | 61440 | 2048 | 408.05 | 128 | 27.68 | 42.919 | 127 | 28.03 | 869655948 |
| 32 | 63488 | 2048 | 398.87 | 128 | 27.89 | 40.892 | 127 | 28.24 | 897844620 |
| 33 | 65536 | 2048 | 398.50 | 128 | 27.64 | 43.172 | 127 | 28.01 | 0 |