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

[test]Corrected the Qwen3-Omni-30B-A3B-Instruct accuracy test configuration in nightly tests. (#5195)

Browse Source 
### What this PR does / why we need it?
Corrected the Qwen3-Omni-30B-A3B-Instruct accuracy test configuration in
nightly tests.
link: https://github.com/vllm-project/vllm-ascend/pull/4911

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c
---------
Signed-off-by: MrZ20 <2609716663@qq.com>
Signed-off-by: root <root@LAPTOP-VQKDDVMG.localdomain>
Co-authored-by: root <root@LAPTOP-VQKDDVMG.localdomain>
This commit is contained in:
SILONG ZENG
2025-12-23 14:17:27 +08:00
committed by GitHub
parent 29a93daa82
commit fa0c212bfa
2 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
.github/workflows/nightly_test_a2.yaml vendored
View File

@@ -125,7 +125,7 @@ jobs:
- os: linux-aarch64-a2-4 - os: linux-aarch64-a2-4
model_list: model_list:
- Qwen3-Next-80B-A3B-Instruct - Qwen3-Next-80B-A3B-Instruct
- Qwen3-VL-30B-A3B-Instruct - Qwen3-Omni-30B-A3B-Instruct
uses: ./.github/workflows/_e2e_nightly_single_node_models.yaml uses: ./.github/workflows/_e2e_nightly_single_node_models.yaml
with: with:
vllm: v0.13.0 vllm: v0.13.0

2
tests/e2e/models/configs/Qwen3-Omni-30B-A3B-Instruct.yaml
View File

@@ -5,7 +5,7 @@ tasks:
- name: "mmmu_val" - name: "mmmu_val"
metrics: metrics:
- name: "acc,none" - name: "acc,none"
value: 0.52 value: 0.60
max_model_len: 8192 max_model_len: 8192
tensor_parallel_size: 4 tensor_parallel_size: 4
enable_expert_parallel: True enable_expert_parallel: True