Fix inference device placement for quantized and dispatched models #775
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
| name: Minimum version tests | |
| on: | |
| push: | |
| branches: | |
| - main | |
| - "*-release" | |
| pull_request: | |
| branches: | |
| - main | |
| - "*-release" | |
| paths: | |
| - "pyproject.toml" | |
| - "sentence_transformers/**" | |
| - ".github/workflows/min-versions.yml" | |
| workflow_dispatch: | |
| env: | |
| TRANSFORMERS_IS_CI: 1 | |
| HF_HUB_DISABLE_PROGRESS_BARS: 1 | |
| HF_TOKEN: ${{ secrets.HF_TOKEN }} # Read token to avoid rate limits / gated-repo failures when downloading from the Hub | |
| jobs: | |
| test_min_versions: | |
| name: Test minimum dependency versions | |
| runs-on: ubuntu-latest | |
| steps: | |
| - name: Remove unnecessary files | |
| run: | | |
| sudo rm -rf \ | |
| "$AGENT_TOOLSDIRECTORY" \ | |
| /opt/google/chrome \ | |
| /opt/microsoft/msedge \ | |
| /opt/microsoft/powershell \ | |
| /opt/pipx \ | |
| /usr/lib/mono \ | |
| /usr/local/julia* \ | |
| /usr/local/lib/android \ | |
| /usr/local/lib/node_modules \ | |
| /usr/local/share/chromium \ | |
| /usr/local/share/powershell \ | |
| /usr/share/dotnet \ | |
| /usr/share/swift | |
| - name: Checkout code | |
| uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 | |
| - name: Setup Python environment | |
| uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0 | |
| with: | |
| python-version: "3.10" | |
| - name: Install uv | |
| uses: astral-sh/setup-uv@20cfd1bf945f4377ade1205e4dbc17946fc9a30d # v10.0.1 | |
| - name: Install FFmpeg libraries (needed to build av) | |
| run: sudo apt-get update && sudo apt-get install -y libavformat-dev libavcodec-dev libavdevice-dev libavutil-dev libavfilter-dev libswscale-dev libswresample-dev | |
| - name: Install with minimum versions | |
| # Runtime deps at lowest resolution & dev tools at normal resolution | |
| # av>=12 because av==9.2.0 (from transformers[video]) fails to compile with modern Cython | |
| # scipy>=1.7.2 because older scipy has no wheels for Python 3.10+ and fails to build from source | |
| # accelerate>=1.13.0 because older accelerate cannot dispatch streaming (IterableDataset) batches | |
| # with string columns, breaking the streaming trainer tests | |
| run: | | |
| uv pip install '.[train]' 'transformers[vision,audio,video]' 'av>=12.0.0' 'scipy>=1.7.2' 'accelerate>=1.13.0' --resolution lowest-direct --system | |
| uv pip install pytest pytest-cov pytest-env pytest-subtests pytest-xdist peft --system | |
| - name: Show installed versions | |
| run: uv pip list --system | grep -iE 'transformers|huggingface|torch|numpy|scikit|scipy|typing.ext|tqdm|accelerate' | |
| - name: Run unit tests | |
| run: python -m pytest --durations 20 -sv tests/ |