[CI]Add model basic accuracy test(Qwen2.5-0.5B-Instruct)

.github/workflows/actionlint.yml

@@ -1,5 +1,4 @@
 #
-# Adapted from vllm-project/vllm/blob/main/.github
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -13,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# Adapted from vllm-project/vllm/blob/main/.github
 #
 
 name: Lint GitHub Actions workflows

.github/workflows/mypy.yaml

@@ -1,5 +1,4 @@
 #
-# Adapted from vllm-project/vllm/blob/main/.github
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -13,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# Adapted from vllm-project/vllm/blob/main/.github
 #
 
 name: mypy

.github/workflows/ruff.yml

@@ -1,5 +1,4 @@
 #
-# Adapted from vllm-project/vllm/blob/main/.github
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -13,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# Adapted from vllm-project/vllm/blob/main/.github
 #
 
 name: ruff

.github/workflows/shellcheck.yml

@@ -1,5 +1,4 @@
 #
-# Adapted from vllm-project/vllm/blob/main/.github
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -13,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# Adapted from vllm-project/vllm/blob/main/.github
 #
 
 name: Lint shell scripts

.github/workflows/vllm_ascend_test.yaml

@@ -1,6 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -13,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
 #
 
 name: 'e2e test'

.github/workflows/yapf.yml

@@ -1,5 +1,4 @@
 #
-# Adapted from vllm-project/vllm/blob/main/.github
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -13,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# Adapted from vllm-project/vllm/blob/main/.github
 #
 
 name: yapf

Dockerfile

@@ -1,6 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -13,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
 #
 
 FROM quay.io/ascend/cann:8.0.0-910b-ubuntu22.04-py3.10

Dockerfile.openEuler

@@ -1,6 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -13,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
 #
 
 FROM quay.io/ascend/cann:8.0.0-910b-openeuler22.03-py3.10

collect_env.py

@@ -1,5 +1,4 @@
 #
-# Adapted from https://github.com/vllm-project/vllm/blob/main/collect_env.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -13,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# Adapted from https://github.com/vllm-project/vllm/blob/main/collect_env.py
 #
 
 import datetime

docs/source/conf.py

@@ -1,7 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from vllm-project/vllm/docs/source/conf.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -15,6 +13,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from vllm-project/vllm/docs/source/conf.py
 #
 
 # -- Path setup --------------------------------------------------------------

examples/offline_distributed_inference_npu.py

@@ -1,7 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from vllm-project/vllm/examples/offline_inference/basic.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -15,6 +13,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from vllm-project/vllm/examples/offline_inference/basic.py
 #
 
 from vllm import LLM, SamplingParams

examples/offline_inference_audio_language.py

@@ -1,7 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from vllm-project/vllm/examples/offline_inference/audio_language.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -15,6 +13,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from vllm-project/vllm/examples/offline_inference/audio_language.py
 #
 """
 This example shows how to use vLLM for running offline inference 

examples/offline_inference_npu.py

@@ -1,7 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from vllm-project/vllm/examples/offline_inference/basic.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -15,6 +13,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from vllm-project/vllm/examples/offline_inference/basic.py
 #
 
 from vllm import LLM, SamplingParams

format.sh

@@ -2,8 +2,6 @@
 
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from https://github.com/vllm-project/vllm/tree/main/tools
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -17,6 +15,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from https://github.com/vllm-project/vllm/tree/main/tools
 #
 
 # YAPF formatter, adapted from ray and skypilot.

mypy.ini

@@ -9,3 +9,5 @@ ignore_missing_imports = True
 [mypy-transformers.*]
 ignore_missing_imports = True
 
+[mypy-lm_eval.*]
+ignore_missing_imports = True

requirements-dev.txt

@@ -2,3 +2,5 @@
 modelscope
 pytest >= 6.0
 pytest-asyncio
+lm-eval
+ray

setup.py

@@ -1,7 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from https://github.com/vllm-project/vllm/blob/main/setup.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -15,6 +13,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from https://github.com/vllm-project/vllm/blob/main/setup.py
 #
 
 import importlib.util

tests/conftest.py

@@ -1,7 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from vllm-project/vllm/blob/main/tests/conftest.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -15,6 +13,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from vllm-project/vllm/blob/main/tests/conftest.py
 #
 
 import gc

tests/model_utils.py

@@ -1,7 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from vllm-project/vllm/blob/main/tests/models/utils.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -15,6 +13,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from vllm-project/vllm/blob/main/tests/models/utils.py
 #
 
 import warnings

tests/multicard/test_offline_inference_distributed.py

@@ -1,7 +1,5 @@
 #
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from vllm/tests/basic_correctness/test_basic_correctness.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -15,6 +13,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from vllm/tests/basic_correctness/test_basic_correctness.py
 #
 """Compare the short outputs of HF and vLLM when using greedy sampling.
 

tests/ops/test_fused_moe.py

@@ -1,6 +1,4 @@
 # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
-# This file is a part of the vllm-ascend project.
-# Adapted from vllm/tests/kernels/test_moe.py
 # Copyright 2023 The vLLM team.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -15,6 +13,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # SPDX-License-Identifier: Apache-2.0
+# This file is a part of the vllm-ascend project.
+# Adapted from vllm/tests/kernels/test_moe.py
 """Tests for the MOE layers.
 
 Run `pytest tests/ops/test_fused_moe.py`.

tests/ops/test_rotary_embedding.py

@@ -1,8 +1,8 @@
-# Adapted from
-# https://github.com/vllm-project/vllm/blob/main/vllm/tests/kernels/test_rotary_embedding.py
 # Copyright 2023 The vLLM team.
 
 # Copyright (c) Huawei Technologies Co., Ltd. 2024-2025. All rights reserved.
+# Adapted from
+# https://github.com/vllm-project/vllm/blob/main/vllm/tests/kernels/test_rotary_embedding.py
 
 from typing import Optional, Tuple, Union
 

tests/singlecard/test_accuracy.py

@@ -0,0 +1,66 @@
+#
+# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
+# Copyright 2023 The vLLM team.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# This file is a part of the vllm-ascend project.
+# Adapted from vllm-project/blob/main/tests/entrypoints/llm/test_accuracy.py
+#
+
+import gc
+import multiprocessing
+from multiprocessing import Queue
+
+import lm_eval
+import pytest
+import torch
+
+# pre-trained model path on Hugging Face.
+MODEL_NAME = "Qwen/Qwen2.5-0.5B-Instruct"
+# Math reasoning benchmark (Grade School Math 8K).
+TASK = "gsm8k"
+# Answer validation requiring format consistency.
+FILTER = "exact_match,strict-match"
+# 3% relative tolerance for numerical accuracy.
+RTOL = 0.03
+# Baseline accuracy after VLLM optimization.
+EXPECTED_VALUE = 0.316
+
+
+def run_test(queue, more_args=None):
+    model_args = f"pretrained={MODEL_NAME},max_model_len=4096"
+    if more_args is not None:
+        model_args = f"{model_args},{more_args}"
+    results = lm_eval.simple_evaluate(
+        model="vllm",
+        model_args=model_args,
+        tasks=TASK,
+        batch_size="auto",
+    )
+    result = results["results"][TASK][FILTER]
+    print("result:", result)
+    queue.put(result)
+    del results
+    torch.npu.empty_cache()
+    gc.collect()
+
+
+def test_lm_eval_accuracy(monkeypatch: pytest.MonkeyPatch):
+    with monkeypatch.context():
+        result_queue: Queue[float] = multiprocessing.Queue()
+        p = multiprocessing.Process(target=run_test, args=(result_queue, ))
+        p.start()
+        p.join()
+        result = result_queue.get()
+        assert (EXPECTED_VALUE - RTOL < result < EXPECTED_VALUE + RTOL), \
+            f"Expected: {EXPECTED_VALUE}±{RTOL} | Measured: {result}"