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

update vllm pin to 12.27 (#5412)

Browse Source 
### What this PR does / why we need it?
update vllm pin to 12.27
1、Fix Qwen2-MoE shared_expert_gate
：https://github.com/vllm-project/vllm/pull/31339
### Does this PR introduce _any_ user-facing change?

### How was this patch tested?
vLLM version: release/v0.13.0
vLLM main:
5326c89803
Co-authored-by: leo-pony [nengjunma@outlook.com](nengjunma@outlook.com)

---------

Signed-off-by: ZT-AIA <1028681969@qq.com>
Signed-off-by: leo-pony <nengjunma@outlook.com>
Co-authored-by: leo-pony <nengjunma@outlook.com>
This commit is contained in:
ZT-AIA
2025-12-28 00:19:36 +08:00
committed by GitHub
parent 1b5d5abf86
commit 24328aaf00
5 changed files with 7 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
tests/e2e/multicard/test_qwen3_next.py
View File

@@ -70,7 +70,7 @@ def test_qwen3_next_distributed_mp_eager_mtp_similarity_tp4():
"The future of AI is",
]
max_tokens = 20
max_tokens = 15
with VllmRunner(
"Qwen/Qwen3-Next-80B-A3B-Instruct",