update doc (#7046)

Co-authored-by: ximing.wxm <ximing.wxm@antgroup.com>
This commit is contained in:
Ximingwang-09
2025-06-11 10:02:01 +08:00
committed by GitHub
parent 6b12d6a8d5
commit f2a75a66c4

View File

@@ -185,7 +185,7 @@ Please consult the documentation below and [server_args.py](https://github.com/s
| Arguments | Description | Defaults |
|----------|-------------|---------|
| `speculative_draft_model_path` | The draft model path for speculative decoding. | None |
| `speculative_algorithm` | The algorithm for speculative decoding. Currently [EAGLE](https://arxiv.org/html/2406.16858v1) and [EAGLE3](https://arxiv.org/pdf/2503.01840) are supported. Note that the radix cache, chunked prefill, and overlap scheduler are disabled when using eagle speculative decoding. | None |
| `speculative_algorithm` | The algorithm for speculative decoding. Currently [EAGLE](https://arxiv.org/html/2406.16858v1) and [EAGLE3](https://arxiv.org/pdf/2503.01840) are supported. Note that the overlap scheduler is disabled when using eagle speculative decoding. | None |
| `speculative_num_steps` | How many draft passes we run before verifying. | None |
| `speculative_num_draft_tokens` | The number of tokens proposed in a draft. | None |
| `speculative_eagle_topk` | The number of top candidates we keep for verification at each step for [Eagle](https://arxiv.org/html/2406.16858v1). | None |