EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[Bugfix] Fixed an accuracy problem of sp with eagle3 (#5816)

Browse Source 
### What this PR does / why we need it?
Fixed an accuracy problem when using eagle3 with sp.

The problem is described in
https://github.com/vllm-project/vllm-ascend/issues/5825.

It also adds a much more precise way to determine whether drafter should
use `sp` or not.

Also, it changes the `eager` of drafter to be a real `eager` in frontend
to avoid a `fx-graph` problem.

### Does this PR introduce _any_ user-facing change?

N/A

### How was this patch tested?

For simpilicity, we test it as in
https://github.com/vllm-project/vllm-ascend/issues/5825.

And we get the same result of `eagle3` with `sp` disabled.

```text
--------------------------------------------------
total_num_output_tokens: 1000
num_drafts: 437
num_draft_tokens: 1311
num_accepted_tokens: 564
mean acceptance length: 2.29
--------------------------------------------------
acceptance at token 0: 0.62
acceptance at token 1: 0.40
acceptance at token 2: 0.27
acceptance at token 3: 0.00
acceptance at token 4: 0.00
acceptance at token 5: 0.00
```

* vLLM version: v0.13.0
* vLLM main:
2f4e6548ef

Signed-off-by: drslark <slarksblood@qq.com>
This commit is contained in:
drslark
2026-01-14 09:00:37 +08:00
committed by GitHub
parent e1bed43cff
commit 48ec97821a
7 changed files with 246 additions and 141 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
vllm_ascend/utils.py
View File

@@ -62,6 +62,7 @@ _ASCEND_CUSTOMOP_IS_REIGISTERED = False
_DEFAULT_BUFFER_SIZE = 200
_MIN_DP_BUFFER_SIZE = 50
_IS_MOE_MODEL = None
_IS_DRAFTER_MOE_MODEL = None
_IS_VL_MODEL = None
_ENABLE_SP = None
_HAS_LAYER_IDX = None
@@ -842,6 +843,16 @@ def is_moe_model(vllm_config: VllmConfig):
return _IS_MOE_MODEL
def is_drafter_moe_model(vllm_config: VllmConfig):
"""Checks if the drafter model is a MoE model by config"""
global _IS_DRAFTER_MOE_MODEL
if _IS_DRAFTER_MOE_MODEL is None:
model_configs = vllm_config.speculative_config.draft_model_config.hf_text_config \
.to_dict()
_IS_DRAFTER_MOE_MODEL = _is_contain_expert(model_configs)
return _IS_DRAFTER_MOE_MODEL
def speculative_enable_dispatch_gmm_combine_decode(
vllm_config: VllmConfig) -> bool:
if vllm_config.speculative_config is None: