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

[0.11.0] [Cherry-pick #4058] Fixes Qwen3-Next enable nz accuracy problem (#4056)

Browse Source 
### What this PR does / why we need it?
- Fixes Qwen3-Next enable nz accuracy problem

---------

Signed-off-by: wxsIcey <1790571317@qq.com>
Signed-off-by: Icey <1790571317@qq.com>
This commit is contained in:
Icey
2025-11-10 20:56:39 +08:00
committed by GitHub
parent ebd45b6596
commit c5fe179cef
7 changed files with 37 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
vllm_ascend/worker/worker_v1.py
View File

@@ -81,6 +81,7 @@ class NPUWorker(WorkerBase):
# register patch for vllm
from vllm_ascend.utils import adapt_patch
adapt_patch()
is_enable_nz(vllm_config)
# Register ops when worker init.
from vllm_ascend import ops
ops.register_dummy_fusion_op()