title: the acceptance length on answers of about 2,000 tokens, both engines
source: manuals/vllm-101/capture/gpu/out-l4/eagle3-long-outputs.txt
lines: 8-15
lang: text
note: Read the acceptance per draft position, then the comparison with short answers under the table.
---
engine          dataset    mean acceptance length   accepted per draft position 1 / 2 / 3
SGLang EAGLE3   GSM8K      2.68
SGLang EAGLE3   ShareGPT   2.32
vLLM EAGLE3     GSM8K      2.69                     0.77 / 0.54 / 0.38
vLLM EAGLE3     ShareGPT   2.35                     0.67 / 0.42 / 0.26

Short outputs (128 tokens) in the same session on SGLang: GSM8K 2.69 to 2.78, ShareGPT 2.33 to 2.43.
Acceptance did not fall from 128 tokens to about 2,000 tokens on these datasets.
