[CI] Add accuracy CI #16
Workflow file for this run
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| # | |
| # Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved. | |
| # | |
| # 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. | |
| # | |
| name: Benchmarks / accuracy-ci | |
| on: | |
| pull_request: | |
| branches: | |
| - 'main' | |
| - '*-dev' | |
| paths: | |
| - '.github/workflows/accuracy_ci.yaml' | |
| - 'tests/e2e/models/**' | |
| schedule: | |
| - cron: '0 */6 * * *' | |
| # Bash shells do not use ~/.profile or ~/.bashrc so these shells need to be explicitly | |
| # declared as "shell: bash -el {0}" on steps that need to be properly activated. | |
| # It's used to activate ascend-toolkit environment variables. | |
| defaults: | |
| run: | |
| shell: bash -el {0} | |
| # only cancel in-progress runs of the same workflow | |
| concurrency: | |
| group: ${{ github.workflow }}-${{ github.ref }} | |
| cancel-in-progress: true | |
| jobs: | |
| prepare_matrix: | |
| runs-on: ubuntu-latest | |
| outputs: | |
| matrix: ${{ steps.set_matrix.outputs.matrix }} | |
| steps: | |
| - name: Checkout repository | |
| uses: actions/checkout@v4 | |
| with: | |
| fetch-depth: 0 | |
| - name: Install dependencies | |
| run: | | |
| sudo apt-get update | |
| sudo apt-get install -y jq | |
| pip install yq | |
| - name: Set mappings | |
| run: | | |
| mappings_json='{}' | |
| for config_file in tests/e2e/models/configs/*.yaml; do | |
| echo "Processing config file: $config_file" | |
| full_model_name=$(yq -r '.model_name' "$config_file" | tr -d '"') | |
| model_name=$(echo "$full_model_name" | awk -F'/' '{print $NF}') | |
| runner=$(yq -r '.runner' "$config_file" | tr -d '"') | |
| echo "Raw model_name: $full_model_name" | |
| echo "Extracted model_name: $model_name" | |
| echo "Runner: $runner" | |
| mappings_json=$(echo "$mappings_json" | jq --arg key "$model_name" --arg value "$runner" '. + {($key): $value}') | |
| done | |
| compact_json=$(echo "$mappings_json" | jq -c .) | |
| echo "Generated mappings:" | |
| echo "$mappings_json" | |
| echo "MAPPINGS_JSON=$compact_json" >> $GITHUB_ENV | |
| - name: Prepare matrix | |
| id: set_matrix | |
| shell: bash | |
| env: | |
| GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} | |
| run: | | |
| set -e | |
| event="${GITHUB_EVENT_NAME}" | |
| matrix="[]" | |
| if [[ "$event" == "schedule" ]]; then | |
| matrix=$(echo "$MAPPINGS_JSON" | jq '[to_entries[] | {model_name: .key, runner: .value}]') | |
| else | |
| pr_number=$(jq -r '.pull_request.number // empty' "$GITHUB_EVENT_PATH" || true) | |
| changed_files="" | |
| echo "PR detected: #$pr_number — fetching changed files via API" | |
| changed_files=$(curl -s -H "Authorization: Bearer $GITHUB_TOKEN" \ | |
| "$GITHUB_API_URL/repos/$GITHUB_REPOSITORY/pulls/$pr_number/files?per_page=100" \ | |
| | jq -r '.[].filename' || true) | |
| echo "Changed files:" | |
| echo "$changed_files" | |
| run_all=false | |
| for f in $changed_files; do | |
| if [[ "$f" != tests/e2e/models/configs/* ]]; then | |
| echo "Non-config file changed: $f, will run all models" | |
| run_all=true | |
| break | |
| fi | |
| done | |
| if [[ "$run_all" == "true" ]]; then | |
| matrix=$(echo "$MAPPINGS_JSON" | jq '[to_entries[] | {model_name: .key, runner: .value}]') | |
| else | |
| for f in $changed_files; do | |
| if [[ "$f" == tests/e2e/models/configs/*.yaml ]]; then | |
| name=$(basename "$f" .yaml) | |
| runner=$(echo "$MAPPINGS_JSON" | jq -r --arg key "$name" '.[$key] // empty') | |
| if [[ -n "$runner" ]]; then | |
| matrix=$(echo "$matrix" | jq --arg model "$name" --arg runner "$runner" '. += [{"model_name":$model, "runner":$runner}]') | |
| else | |
| echo "Config $name not found in mappings; skipping." | |
| fi | |
| fi | |
| done | |
| fi | |
| fi | |
| echo "Generated matrix (raw): $matrix" | |
| compact_matrix=$(echo "$matrix" | jq -c 'if type=="array" then . else [] end') | |
| echo "matrix=$compact_matrix" >> $GITHUB_OUTPUT | |
| echo "Final matrix output: $compact_matrix" | |
| accuracy_tests: | |
| needs: prepare_matrix | |
| runs-on: ${{ matrix.runner }} | |
| strategy: | |
| matrix: | |
| include: ${{ fromJson(needs.prepare_matrix.outputs.matrix) }} | |
| fail-fast: false | |
| name: ${{ matrix.model_name }} accuracy | |
| container: | |
| image: >- | |
| ${{ | |
| contains(matrix.runner, '310p') | |
| && 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/cann:8.2.rc1-310p-ubuntu22.04-py3.11' | |
| || 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/cann:8.2.rc1-910b-ubuntu22.04-py3.11' | |
| }} | |
| env: | |
| VLLM_USE_MODELSCOPE: True | |
| steps: | |
| - name: Checkout repository | |
| uses: actions/checkout@v4 | |
| - name: Set model name as output | |
| id: set_output | |
| run: | | |
| echo "model_name=${{ matrix.model_name }}" >> $GITHUB_OUTPUT | |
| - name: Config mirrors | |
| run: | | |
| sed -Ei 's@(ports|archive).ubuntu.com@cache-service.nginx-pypi-cache.svc.cluster.local:8081@g' /etc/apt/sources.list | |
| pip config set global.index-url http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple | |
| pip config set global.trusted-host cache-service.nginx-pypi-cache.svc.cluster.local | |
| apt-get update -y | |
| apt install git -y | |
| - name: Install system dependencies | |
| run: | | |
| apt-get -y install `cat packages.txt` | |
| apt-get -y install gcc g++ cmake libnuma-dev | |
| - name: Checkout vllm-project/vllm repo | |
| uses: actions/checkout@v4 | |
| with: | |
| repository: vllm-project/vllm | |
| ref: v0.10.2 | |
| path: ./vllm-empty | |
| - name: Install vllm-project/vllm from source | |
| working-directory: ./vllm-empty | |
| run: | | |
| VLLM_TARGET_DEVICE=empty pip install -e . | |
| - name: Checkout vllm-project/vllm-ascend repo | |
| uses: actions/checkout@v4 | |
| with: | |
| repository: vllm-project/vllm-ascend | |
| path: ./vllm-ascend | |
| - name: Install vllm-project/vllm-ascend | |
| working-directory: ./vllm-ascend | |
| env: | |
| PIP_EXTRA_INDEX_URL: https://mirrors.huaweicloud.com/ascend/repos/pypi | |
| run: | | |
| runner="${{ matrix.runner }}" | |
| # If runner indicates 310p, set SOC_VERSION and LD_LIBRARY_PATH before install | |
| if [[ "$runner" == *310p* ]]; then | |
| export SOC_VERSION=ASCEND310P3 | |
| fi | |
| pip install -r requirements-dev.txt | |
| pip install -v -e . | |
| - name: Get vLLM commit hash and URL | |
| working-directory: ./vllm-empty | |
| run: | | |
| VLLM_COMMIT=$(git rev-parse --short=7 HEAD) | |
| echo "VLLM_COMMIT=$VLLM_COMMIT" >> $GITHUB_ENV | |
| - name: Get vLLM-Ascend commit hash and URL | |
| working-directory: ./vllm-ascend | |
| run: | | |
| VLLM_ASCEND_COMMIT=$(git rev-parse --short=7 HEAD) | |
| echo "VLLM_ASCEND_COMMIT=$VLLM_ASCEND_COMMIT" >> $GITHUB_ENV | |
| - name: Collect version info | |
| run: | | |
| for dir in /usr/local/Ascend/ascend-toolkit/*; do | |
| dname=$(basename "$dir") | |
| if [ "$dname" != "latest" ]; then | |
| TOOLKIT_DIR="$dname" | |
| break | |
| fi | |
| done | |
| INFO_FILE="/usr/local/Ascend/ascend-toolkit/${TOOLKIT_DIR}/$(uname -i)-linux/ascend_toolkit_install.info" | |
| GHA_CANN_VERSION=$(grep "version=" "$INFO_FILE" \ | |
| | head -n1 \ | |
| | cut -d'=' -f2 \ | |
| | tr -d '"') | |
| { | |
| echo "GHA_CANN_VERSION=$GHA_CANN_VERSION" | |
| pip show torch | grep "Version:" | awk '{print "GHA_TORCH_VERSION="$2}' | |
| pip show torch_npu | grep "Version:" | awk '{print "GHA_TORCH_NPU_VERSION="$2}' | |
| pip show vllm | grep "Version:" | awk '{print "GHA_VLLM_VERSION="$2}' | sed 's/+.*//' | |
| } >> "$GITHUB_ENV" | |
| - name: Run accuracy test | |
| id: report | |
| env: | |
| VLLM_WORKER_MULTIPROC_METHOD: spawn | |
| VLLM_USE_MODELSCOPE: True | |
| VLLM_VERSION: ${{ env.GHA_VLLM_VERSION }} | |
| VLLM_COMMIT: ${{ env.VLLM_COMMIT }} | |
| VLLM_ASCEND_VERSION: ${{ env.GHA_VLLM_ASCEND_VERSION || github.ref }} | |
| VLLM_ASCEND_COMMIT: ${{ env.VLLM_ASCEND_COMMIT }} | |
| CANN_VERSION: ${{ env.GHA_CANN_VERSION }} | |
| TORCH_VERSION: ${{ env.GHA_TORCH_VERSION }} | |
| TORCH_NPU_VERSION: ${{ env.GHA_TORCH_NPU_VERSION }} | |
| run: | | |
| model_base_name=$(basename ${{ matrix.model_name }}) | |
| markdown_name="${model_base_name}" | |
| echo "markdown_name=$markdown_name" >> $GITHUB_OUTPUT | |
| mkdir -p ./benchmarks/accuracy | |
| pytest -sv ./tests/e2e/models/test_lm_eval_correctness.py \ | |
| --config ./tests/e2e/models/configs/${{ matrix.model_name }}.yaml | |
| - name: Generate step summary | |
| if: ${{ always() }} | |
| run: | | |
| cat ./benchmarks/accuracy/${{ steps.report.outputs.markdown_name }}.md >> $GITHUB_STEP_SUMMARY | |
| - name: Sanitize version string for artifact naming | |
| run: | | |
| SAFE_VLLM_ASCEND_VERSION="${GHA_VLLM_ASCEND_VERSION//\//-}" | |
| echo "SAFE_VLLM_ASCEND_VERSION=$SAFE_VLLM_ASCEND_VERSION" >> "$GITHUB_ENV" | |
| - name: Upload Report | |
| uses: actions/upload-artifact@v4 | |
| with: | |
| name: "report-${{ env.SAFE_VLLM_ASCEND_VERSION }}-${{ steps.report.outputs.markdown_name }}" | |
| path: ./benchmarks/accuracy/${{ steps.report.outputs.markdown_name }}.md | |
| if-no-files-found: warn | |
| retention-days: 90 | |
| overwrite: true |