From 49e6983b3be77873cd21ddd9b4cf3deddf65e0ea Mon Sep 17 00:00:00 2001
From: zhangxinyuehfad <59153331+zhangxinyuehfad@users.noreply.github.com>
Date: Tue, 4 Nov 2025 18:56:31 +0800
Subject: [PATCH] [Test] Add accuracy test for qwen3-30b-a3b-w8a8 (#3807)

### What this PR does / why we need it?
Add accuracy test for qwen3-30b-a3b-w8a8
This PR depends on https://github.com/vllm-project/vllm-ascend/pull/3799

### How was this patch tested?
qwen3-30b-a3b-w8a8 accuarcy test ok:

https://github.com/vllm-project/vllm-ascend/actions/runs/19062045267/job/54443732877?pr=3807
- vLLM version: v0.11.0
- vLLM main:
https://github.com/vllm-project/vllm/commit/83f478bb19489b41e9d208b47b4bb5a95ac171ac

Signed-off-by: hfadzxy <starmoon_zhang@163.com>
---
 .github/workflows/accuracy_test.yaml             |  2 ++
 tests/e2e/models/configs/Qwen3-30B-A3B-W8A8.yaml | 16 ++++++++++++++++
 2 files changed, 18 insertions(+)
 create mode 100644 tests/e2e/models/configs/Qwen3-30B-A3B-W8A8.yaml

diff --git a/.github/workflows/accuracy_test.yaml b/.github/workflows/accuracy_test.yaml
index 5621b615..7a1b5c39 100644
--- a/.github/workflows/accuracy_test.yaml
+++ b/.github/workflows/accuracy_test.yaml
@@ -68,6 +68,8 @@ jobs:
             model_name: Qwen2.5-Omni-7B
           - runner: a2-1
             model_name: Meta-Llama-3.1-8B-Instruct
+          - runner: a2-4
+            model_name: Qwen3-30B-A3B-W8A8
       fail-fast: false
     # test will be triggered when tag 'accuracy-test' & 'ready-for-test'
     if:  >-
diff --git a/tests/e2e/models/configs/Qwen3-30B-A3B-W8A8.yaml b/tests/e2e/models/configs/Qwen3-30B-A3B-W8A8.yaml
new file mode 100644
index 00000000..6379c45d
--- /dev/null
+++ b/tests/e2e/models/configs/Qwen3-30B-A3B-W8A8.yaml
@@ -0,0 +1,16 @@
+model_name: "vllm-ascend/Qwen3-30B-A3B-W8A8"
+hardware: "Atlas A2 Series"
+tasks:
+- name: "gsm8k"
+  metrics:
+  - name: "exact_match,strict-match"
+    value: 0.9
+  - name: "exact_match,flexible-extract"
+    value: 0.8
+num_fewshot: 5
+gpu_memory_utilization: 0.7
+enable_expert_parallel: True
+tensor_parallel_size: 2
+apply_chat_template: False
+fewshot_as_multiturn: False
+quantization: ascend