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

[BugFix][main] Adapted Qwen3-Next-MTP to chunked prefill (#4770)

Browse Source 
### What this PR does / why we need it?
The pad `-1` modification is from
https://github.com/vllm-project/vllm/pull/25743.

It still has bugs for batched chunked prefill.

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

Signed-off-by: drslark <slarksblood@qq.com>
Co-authored-by: Mengqing Cao <cmq0113@163.com>
This commit is contained in:
drslark
2025-12-10 22:54:24 +08:00
committed by GitHub
parent 490ddf536f
commit 0fb1dc43a1
8 changed files with 646 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
vllm_ascend/patch/worker/__init__.py
View File

@@ -32,3 +32,4 @@ import vllm_ascend.patch.worker.patch_qwen2_5_vl # noqa
import vllm_ascend.patch.worker.patch_qwen2_5_omni # noqa
import vllm_ascend.patch.worker.patch_qwen3_vl # noqa
import vllm_ascend.patch.worker.patch_rope # noqa
import vllm_ascend.patch.worker.patch_qwen3_next_mtp # noqa