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

[Bugfix] Fix setting of speculative_config.enforce_eager for dsv32 (#5945)

Browse Source 
### What this PR does / why we need it?
This PR aims to fix setting of `speculative_config.enforce_eager` in
deepseek v3.2 mtp. The point is that, vllm sets
`speculative_config.enforce_eager` as True if using deepseek_v32 with
mtp. Since we support graph mode, we simply ignore it here. However,
this fix will also implicitly ignore user setting of
`speculative_config.enforce_eager`, we need to take care and remove it
once vllm supports this feature.

### Does this PR introduce _any_ user-facing change?
N/A

### How was this patch tested?
by ci

- vLLM version: v0.13.0
- vLLM main:
2c24bc6996

Signed-off-by: Zetong Li <slippersss@126.com>
This commit is contained in:
Zetong Li
2026-01-21 09:24:33 +08:00
committed by GitHub
parent 936d81a258
commit 1ab6cd4935
2 changed files with 16 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm_ascend/spec_decode/mtp_proposer.py
View File

@@ -246,7 +246,7 @@ class MtpProposer(EagleProposer):
-1]:
num_input_tokens = self.vllm_config.pad_for_cudagraph(
num_scheduled_tokens)
elif self.use_aclgraph and num_tokens <= self.runner.cudagraph_batch_sizes[
elif self.use_aclgraph and num_tokens <= self.runner.cudagraph_batch_sizes[
-1]:
# Acl graph mode, add padding to the batch size
num_input_tokens = self.vllm_config.pad_for_cudagraph(num_tokens)