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

[BugFix] Fixes Qwen3-Next enable nz accuracy problem (#4058)

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

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


- vLLM version: v0.11.0
- vLLM main:
83f478bb19

---------

Signed-off-by: Icey <1790571317@qq.com>
Signed-off-by: wxsIcey <1790571317@qq.com>
This commit is contained in:
Icey
2025-11-10 20:54:57 +08:00
committed by GitHub
parent e6625bb582
commit e04a87f4be
8 changed files with 36 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
tests/e2e/multicard/test_qwen3_next.py
View File

@@ -20,17 +20,9 @@
Run `pytest tests/e2e/multicard/test_qwen3_next.py`.
"""
import os
from unittest.mock import patch
from tests.e2e.conftest import VllmRunner
# NZ will cause precision error in Qwen3-Next
# When it is fixed, this set-up can be removed
_IS_ENABLE_NZ = "VLLM_ASCEND_ENABLE_NZ"
@patch.dict(os.environ, {_IS_ENABLE_NZ: "0"})
def test_models_distributed_Qwen3_NEXT_TP4():
example_prompts = [
"Hello, my name is",
@@ -46,7 +38,6 @@ def test_models_distributed_Qwen3_NEXT_TP4():
del vllm_model
@patch.dict(os.environ, {_IS_ENABLE_NZ: "0"})
def test_models_distributed_Qwen3_NEXT_TP4_FULL_DECODE_ONLY():
example_prompts = [
"Hello, my name is",
@@ -66,7 +57,6 @@ def test_models_distributed_Qwen3_NEXT_TP4_FULL_DECODE_ONLY():
del vllm_model
@patch.dict(os.environ, {_IS_ENABLE_NZ: "0"})
def test_models_distributed_Qwen3_NEXT_MTP_TP4_SIMILARITY():
example_prompts = [
"Hello, my name is",