update acc

Signed-off-by: jiangyunfan1 <jiangyunfan1@h-partners.com>

Author: jiangyunfan1

.github/workflows/vllm_ascend_test_nightly_a3.yaml

@@ -141,3 +141,20 @@ jobs:
       image: 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:nightly-a3'
       tests: ${{ matrix.test_config.tests }}
       name: ${{ matrix.test_config.name }}
+
+  single-node-tests2:
+    name: single-node
+    strategy:
+      fail-fast: false
+      matrix:
+        test_config:
+          - name: deepseek-r1-w8a8-mtpx
+            os: linux-aarch64-a3-16
+            tests: tests/e2e/nightly/features/test_mtpx_deepseek_r1_0528_w8a8.py
+    uses: ./.github/workflows/_e2e_nightly_single_node.yaml
+    with:
+      vllm: v0.11.0
+      runner: ${{ matrix.test_config.os }}
+      image: 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:nightly-a3'
+      tests: ${{ matrix.test_config.tests }}
+      name: ${{ matrix.test_config.name }}

tests/e2e/nightly/features/test_mtpx_deepseek_r1_0528_w8a8.py

@@ -38,17 +38,30 @@
     "max_tokens": 10,
 }
 
-aisbench_cases = [{
+aisbench_gsm8k = [{
+    "case_type": "accuracy",
+    "dataset_path": "vllm-ascend/gsm8k-lite",
+    "request_conf": "vllm_api_general_chat",
+    "dataset_conf": "gsm8k/gsm8k_gen_0_shot_noncot_chat_prompt",
+    "max_out_len": 32768,
+    "batch_size": 32,
+    "baseline": 95,
+    "threshold": 5
+}]
+
+aisbench_aime = [{
     "case_type": "accuracy",
     "dataset_path": "vllm-ascend/aime2024",
     "request_conf": "vllm_api_general_chat",
     "dataset_conf": "aime2024/aime2024_gen_0_shot_chat_prompt",
     "max_out_len": 32768,
     "batch_size": 32,
-    "baseline": 80,
+    "baseline": 86.67,
     "threshold": 7
 }]
 
+aisbench_case_dict = {"mtp2": aisbench_gsm8k, "mtp3": aisbench_aime}
+
 
 @pytest.mark.asyncio
 @pytest.mark.parametrize("model", MODELS)
@@ -132,6 +145,7 @@ async def test_models(model: str, mode: str) -> None:
         assert choices[0].text, "empty response"
         print(choices)
         # aisbench test
+        aisbench_cases = aisbench_case_dict[mode]
         run_aisbench_cases(model,
                            port,
                            aisbench_cases,