refactor: extract common build steps into reusable composite action

- Create .github/actions/build-and-upload composite action for shared build logic
- Consolidate Python/uv setup, CUDA setup, wheel building, testing, and upload steps
- Update _build_linux.yml to use new composite action (reduced from 136 to 69 lines)
- Update _build_linux_arm_self_host.yml to use composite action with cleanup enabled
- Update _build_linux_self_host.yml for both container and no-container jobs
- Update _build_manylinux_self_host.yml to use composite action
- Reduce code duplication by ~200+ lines across all workflow files

.github/actions/build-and-upload/action.yml

@@ -0,0 +1,127 @@
+name: "Build and Upload Flash Attention Wheels"
+description: "Build, test, and upload Flash Attention wheels with optional manylinux support"
+
+inputs:
+  flash-attn-version:
+    description: "Flash-Attention version"
+    required: true
+  python-version:
+    description: "Python version"
+    required: true
+  torch-version:
+    description: "PyTorch version"
+    required: true
+  cuda-version:
+    description: "CUDA version"
+    required: true
+  is-upload:
+    description: "Whether to upload the release asset"
+    required: false
+    default: "true"
+  cleanup:
+    description: "Whether to clean up Python installation"
+    required: false
+    default: "false"
+
+outputs:
+  wheel-path:
+    description: "Path to the built wheel"
+    value: ${{ steps.build_wheels.outputs.WHEEL_PATH }}
+  wheel-path-manylinux:
+    description: "Path to the manylinux wheel"
+    value: ${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}
+
+runs:
+  using: "composite"
+  steps:
+    - name: Install uv
+      uses: astral-sh/setup-uv@v7
+
+    - name: Install Python
+      shell: bash
+      run: |
+        uv venv -p ${{ inputs.python-version }}
+        uv pip install -U pip setuptools==75.8.0 wheel packaging psutil auditwheel
+        current_dir=$(pwd)
+        echo "$current_dir/.venv/bin" >> $GITHUB_PATH
+
+    - name: Setup CUDA
+      uses: mjun0812/setup-cuda@v1
+      with:
+        version: ${{ inputs.cuda-version }}
+
+    - name: Build wheels
+      id: build_wheels
+      shell: bash
+      run: |
+        chmod +x build_linux.sh
+        ./build_linux.sh ${{ inputs.flash-attn-version }} ${{ inputs.python-version }} ${{ inputs.torch-version }} ${{ inputs.cuda-version }}
+        wheel_path=$(ls flash-attention/dist/*.whl | head -n 1)
+        echo "WHEEL_PATH=$wheel_path" >> $GITHUB_OUTPUT
+
+    - name: Install Test
+      shell: bash
+      run: |
+        pip uninstall -y flash-attn > /dev/null 2>&1
+        pip install --no-cache-dir ${{ steps.build_wheels.outputs.WHEEL_PATH }}
+        python -c "import flash_attn; print(flash_attn.__version__)"
+
+    - name: Upload Release Asset
+      if: ${{ inputs.is-upload == 'true' }}
+      shell: bash
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
+      run: |
+        tag_name=${{ github.ref_name }}
+        wheel_path="${{ steps.build_wheels.outputs.WHEEL_PATH }}"
+        # Check if the file exists
+        if [ ! -f "$wheel_path" ]; then
+          echo "Error: Wheel file not found at $wheel_path"
+          exit 1
+        fi
+        # Upload the release asset using GitHub CLI
+        gh release upload "$tag_name" "$wheel_path" --clobber
+
+    - name: Apply auditwheel repair
+      continue-on-error: true
+      id: auditwheel_repair
+      shell: bash
+      run: |
+        auditwheel show ${{ steps.build_wheels.outputs.WHEEL_PATH }}
+        auditwheel repair \
+          --exclude libc10* --exclude libtorch* --exclude libcu* --exclude libnv* --exclude 'libtorch*' \
+          ${{ steps.build_wheels.outputs.WHEEL_PATH }} -w flash-attention/dist_manylinux
+        wheel_path_manylinux=$(ls flash-attention/dist_manylinux/*manylinux*.whl 2>/dev/null | head -n 1 || echo "")
+        echo "WHEEL_PATH_MANYLINUX=$wheel_path_manylinux" >> $GITHUB_OUTPUT
+
+    - name: Test manylinux wheel
+      continue-on-error: true
+      shell: bash
+      run: |
+        if [ -n "${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}" ] && [ -f "${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}" ]; then
+          pip uninstall -y flash-attn > /dev/null 2>&1
+          pip install --no-cache-dir ${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}
+          python -c "import flash_attn; print(flash_attn.__version__)"
+        fi
+
+    - name: Upload manylinux wheel
+      if: ${{ inputs.is-upload == 'true' }}
+      continue-on-error: true
+      shell: bash
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
+      run: |
+        tag_name=${{ github.ref_name }}
+        wheel_path_manylinux="${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}"
+        if [ -n "$wheel_path_manylinux" ] && [ -f "$wheel_path_manylinux" ]; then
+          # Upload the release asset using GitHub CLI
+          gh release upload "$tag_name" "$wheel_path_manylinux" --clobber
+        else
+          echo "Skipping manylinux wheel upload: file not found"
+        fi
+
+    - name: Clean up
+      if: ${{ inputs.cleanup == 'true' }}
+      shell: bash
+      run: |
+        rm -rf /opt/hostedtoolcache/Python

.github/workflows/_build_linux.yml

@@ -58,78 +58,11 @@ jobs:
         run: |
           sudo apt install -y ninja-build clang time patchelf
 
-      # Install Python using uv because setup-python needs newer version of glibc
+      - name: Build and upload wheels
-      - name: Install uv
+        uses: ./.github/actions/build-and-upload
-        uses: astral-sh/setup-uv@v7
-      - name: Install Python
-        shell: bash
-        run: |
-          uv venv -p ${{ inputs.python-version }}
-          uv pip install -U pip setuptools==75.8.0 wheel packaging psutil auditwheel
-          current_dir=$(pwd)
-          echo "$current_dir/.venv/bin" >> $GITHUB_PATH
-
-      - uses: mjun0812/setup-cuda@v1
         with:
-          version: ${{ inputs.cuda-version }}
+          flash-attn-version: ${{ inputs.flash-attn-version }}
-
+          python-version: ${{ inputs.python-version }}
-      - name: Build wheels
+          torch-version: ${{ inputs.torch-version }}
-        id: build_wheels
+          cuda-version: ${{ inputs.cuda-version }}
-        run: |
+          is-upload: ${{ inputs.is-upload }}
-          chmod +x build_linux.sh
-          ./build_linux.sh ${{ inputs.flash-attn-version }} ${{ inputs.python-version }} ${{ inputs.torch-version }} ${{ inputs.cuda-version }}
-          wheel_path=$(ls flash-attention/dist/*.whl | head -n 1)
-          echo "WHEEL_PATH=$wheel_path" >> $GITHUB_OUTPUT
-
-      - name: Install Test
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload Release Asset
-        if: ${{ inputs.is-upload }}
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path="${{ steps.build_wheels.outputs.WHEEL_PATH }}"
-          # Check if the file exists
-          if [ ! -f "$wheel_path" ]; then
-            echo "Error: Wheel file not found at $wheel_path"
-            exit 1
-          fi
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path" --clobber
-
-      - name: Apply auditwheel repair
-        continue-on-error: true
-        id: auditwheel_repair
-        run: |
-          auditwheel show ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          auditwheel repair \
-            --exclude libc10* --exclude libtorch* --exclude libcu* --exclude libnv* --exclude 'libtorch*' \
-            ${{ steps.build_wheels.outputs.WHEEL_PATH }} -w flash-attention/dist_manylinux
-          wheel_path_manylinux=$(ls flash-attention/dist_manylinux/*manylinux*.whl | head -n 1)
-          echo "WHEEL_PATH_MANYLINUX=$wheel_path_manylinux" >> $GITHUB_OUTPUT
-
-      - name: Test manylinux wheel
-        continue-on-error: true
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload manylinux wheel
-        continue-on-error: true
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path_manylinux="${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}"
-          if [ ! -f "$wheel_path_manylinux" ]; then
-            echo "Error: Wheel file not found at $wheel_path_manylinux"
-            exit 1
-          fi
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path_manylinux" --clobber

.github/workflows/_build_linux_arm_self_host.yml

@@ -95,90 +95,12 @@ jobs:
         run: |
           git config --global --add safe.directory $(pwd)
 
-      # Install Python using uv because setup-python needs newer version of glibc
+      - name: Build and upload wheels
-      - name: Install uv
+        uses: ./.github/actions/build-and-upload
-        uses: astral-sh/setup-uv@v7
-      - name: Install Python
-        shell: bash
-        run: |
-          uv venv -p ${{ inputs.python-version }}
-          uv pip install -U pip setuptools==75.8.0 wheel packaging psutil auditwheel
-          current_dir=$(pwd)
-          echo "$current_dir/.venv/bin" >> $GITHUB_PATH
-
-      - uses: mjun0812/setup-cuda@v1
         with:
-          version: ${{ inputs.cuda-version }}
+          flash-attn-version: ${{ inputs.flash-attn-version }}
-
+          python-version: ${{ inputs.python-version }}
-      - name: Build wheels
+          torch-version: ${{ inputs.torch-version }}
-        timeout-minutes: 2160
+          cuda-version: ${{ inputs.cuda-version }}
-        id: build_wheels
+          is-upload: ${{ inputs.is-upload }}
-        shell: bash
+          cleanup: "true"
-        run: |
-          chmod +x build_linux.sh
-          ./build_linux.sh ${{ inputs.flash-attn-version }} ${{ inputs.python-version }} ${{ inputs.torch-version }} ${{ inputs.cuda-version }}
-          wheel_path=$(ls flash-attention/dist/*.whl | head -n 1)
-          echo "WHEEL_PATH=$wheel_path" >> $GITHUB_OUTPUT
-
-      - name: Install Test
-        shell: bash
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload Release Asset
-        if: ${{ inputs.is-upload }}
-        shell: bash
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path="${{ steps.build_wheels.outputs.WHEEL_PATH }}"
-
-          # Check if the file exists
-          if [ ! -f "$wheel_path" ]; then
-            echo "Error: Wheel file not found at $wheel_path"
-            exit 1
-          fi
-
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path" --clobber
-
-      - name: Apply auditwheel repair
-        continue-on-error: true
-        id: auditwheel_repair
-        run: |
-          auditwheel show ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          auditwheel repair \
-            --exclude libc10* --exclude libtorch* --exclude libcu* --exclude libnv* --exclude 'libtorch*' \
-            ${{ steps.build_wheels.outputs.WHEEL_PATH }} -w flash-attention/dist_manylinux
-          wheel_path_manylinux=$(ls flash-attention/dist_manylinux/*manylinux*.whl | head -n 1)
-          echo "WHEEL_PATH_MANYLINUX=$wheel_path_manylinux" >> $GITHUB_OUTPUT
-
-      - name: Test manylinux wheel
-        continue-on-error: true
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload manylinux wheel
-        continue-on-error: true
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path_manylinux="${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}"
-          if [ ! -f "$wheel_path_manylinux" ]; then
-            echo "Error: Wheel file not found at $wheel_path_manylinux"
-            exit 1
-          fi
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path_manylinux" --clobber
-
-      - name: Clean up
-        shell: bash
-        if: always()
-        run: |
-          rm -rf /opt/hostedtoolcache/Python

.github/workflows/_build_linux_self_host.yml

@@ -100,93 +100,15 @@ jobs:
         run: |
           git config --global --add safe.directory $(pwd)
 
-      # Install Python using uv because setup-python needs newer version of glibc
+      - name: Build and upload wheels
-      - name: Install uv
+        uses: ./.github/actions/build-and-upload
-        uses: astral-sh/setup-uv@v7
-      - name: Install Python
-        shell: bash
-        run: |
-          uv venv -p ${{ inputs.python-version }}
-          uv pip install -U pip setuptools==75.8.0 wheel packaging psutil auditwheel
-          current_dir=$(pwd)
-          echo "$current_dir/.venv/bin" >> $GITHUB_PATH
-
-      - uses: mjun0812/setup-cuda@v1
         with:
-          version: ${{ inputs.cuda-version }}
+          flash-attn-version: ${{ inputs.flash-attn-version }}
-
+          python-version: ${{ inputs.python-version }}
-      - name: Build wheels
+          torch-version: ${{ inputs.torch-version }}
-        timeout-minutes: 2160
+          cuda-version: ${{ inputs.cuda-version }}
-        id: build_wheels
+          is-upload: ${{ inputs.is-upload }}
-        shell: bash
+          cleanup: "true"
-        run: |
-          chmod +x build_linux.sh
-          ./build_linux.sh ${{ inputs.flash-attn-version }} ${{ inputs.python-version }} ${{ inputs.torch-version }} ${{ inputs.cuda-version }}
-          wheel_path=$(ls flash-attention/dist/*.whl | head -n 1)
-          echo "WHEEL_PATH=$wheel_path" >> $GITHUB_OUTPUT
-
-      - name: Install Test
-        shell: bash
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload Release Asset
-        if: ${{ inputs.is-upload }}
-        shell: bash
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path="${{ steps.build_wheels.outputs.WHEEL_PATH }}"
-
-          # Check if the file exists
-          if [ ! -f "$wheel_path" ]; then
-            echo "Error: Wheel file not found at $wheel_path"
-            exit 1
-          fi
-
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path" --clobber
-
-      - name: Apply auditwheel repair
-        continue-on-error: true
-        id: auditwheel_repair
-        run: |
-          auditwheel show ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          auditwheel repair \
-            --exclude libc10* --exclude libtorch* --exclude libcu* --exclude libnv* --exclude 'libtorch*' \
-            ${{ steps.build_wheels.outputs.WHEEL_PATH }} -w flash-attention/dist_manylinux
-          wheel_path_manylinux=$(ls flash-attention/dist_manylinux/*manylinux*.whl | head -n 1)
-          echo "WHEEL_PATH_MANYLINUX=$wheel_path_manylinux" >> $GITHUB_OUTPUT
-
-      - name: Test manylinux wheel
-        continue-on-error: true
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload manylinux wheel
-        continue-on-error: true
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path_manylinux="${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}"
-          if [ ! -f "$wheel_path_manylinux" ]; then
-            echo "Error: Wheel file not found at $wheel_path_manylinux"
-            exit 1
-          fi
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path_manylinux" --clobber
-
-      - name: Clean up
-        shell: bash
-        if: always()
-        run: |
-          rm -rf /opt/hostedtoolcache/Python
 
   build_wheels_self_hosted_no_container:
     if: ${{ !inputs.use-container }}
@@ -253,90 +175,12 @@ jobs:
         run: |
           git config --global --add safe.directory $(pwd)
 
-      # Install Python using uv because setup-python needs newer version of glibc
+      - name: Build and upload wheels
-      - name: Install uv
+        uses: ./.github/actions/build-and-upload
-        uses: astral-sh/setup-uv@v7
-      - name: Install Python
-        shell: bash
-        run: |
-          uv venv -p ${{ inputs.python-version }}
-          uv pip install -U pip setuptools==75.8.0 wheel packaging psutil auditwheel
-          current_dir=$(pwd)
-          echo "$current_dir/.venv/bin" >> $GITHUB_PATH
-
-      - uses: mjun0812/setup-cuda@v1
         with:
-          version: ${{ inputs.cuda-version }}
+          flash-attn-version: ${{ inputs.flash-attn-version }}
-
+          python-version: ${{ inputs.python-version }}
-      - name: Build wheels
+          torch-version: ${{ inputs.torch-version }}
-        timeout-minutes: 2160
+          cuda-version: ${{ inputs.cuda-version }}
-        id: build_wheels
+          is-upload: ${{ inputs.is-upload }}
-        shell: bash
+          cleanup: "true"
-        run: |
-          chmod +x build_linux.sh
-          ./build_linux.sh ${{ inputs.flash-attn-version }} ${{ inputs.python-version }} ${{ inputs.torch-version }} ${{ inputs.cuda-version }}
-          wheel_path=$(ls flash-attention/dist/*.whl | head -n 1)
-          echo "WHEEL_PATH=$wheel_path" >> $GITHUB_OUTPUT
-
-      - name: Install Test
-        shell: bash
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload Release Asset
-        if: ${{ inputs.is-upload }}
-        shell: bash
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path="${{ steps.build_wheels.outputs.WHEEL_PATH }}"
-
-          # Check if the file exists
-          if [ ! -f "$wheel_path" ]; then
-            echo "Error: Wheel file not found at $wheel_path"
-            exit 1
-          fi
-
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path" --clobber
-
-      - name: Apply auditwheel repair
-        continue-on-error: true
-        id: auditwheel_repair
-        run: |
-          auditwheel show ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          auditwheel repair \
-            --exclude libc10* --exclude libtorch* --exclude libcu* --exclude libnv* --exclude 'libtorch*' \
-            ${{ steps.build_wheels.outputs.WHEEL_PATH }} -w flash-attention/dist_manylinux
-          wheel_path_manylinux=$(ls flash-attention/dist_manylinux/*manylinux*.whl | head -n 1)
-          echo "WHEEL_PATH_MANYLINUX=$wheel_path_manylinux" >> $GITHUB_OUTPUT
-
-      - name: Test manylinux wheel
-        continue-on-error: true
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload manylinux wheel
-        continue-on-error: true
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path_manylinux="${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}"
-          if [ ! -f "$wheel_path_manylinux" ]; then
-            echo "Error: Wheel file not found at $wheel_path_manylinux"
-            exit 1
-          fi
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path_manylinux" --clobber
-
-      - name: Clean up
-        shell: bash
-        if: always()
-        run: |
-          rm -rf /opt/hostedtoolcache/Python

.github/workflows/_build_manylinux_self_host.yml

@@ -90,93 +90,12 @@ jobs:
         run: |
           git config --global --add safe.directory $(pwd)
 
-      # Install Python using uv because setup-python needs newer version of glibc
+      - name: Build and upload wheels
-      - name: Install uv
+        uses: ./.github/actions/build-and-upload
-        uses: astral-sh/setup-uv@v7
-      - name: Install Python
-        shell: bash
-        run: |
-          uv venv -p ${{ inputs.python-version }}
-          uv pip install -U pip setuptools==75.8.0 wheel packaging psutil auditwheel
-          current_dir=$(pwd)
-          echo "$current_dir/.venv/bin" >> $GITHUB_PATH
-
-      - uses: mjun0812/setup-cuda@v1
         with:
-          version: ${{ inputs.cuda-version }}
+          flash-attn-version: ${{ inputs.flash-attn-version }}
-
+          python-version: ${{ inputs.python-version }}
-      - name: Build wheels
+          torch-version: ${{ inputs.torch-version }}
-        timeout-minutes: 2160
+          cuda-version: ${{ inputs.cuda-version }}
-        id: build_wheels
+          is-upload: ${{ inputs.is-upload }}
-        shell: bash
+          cleanup: "true"
-        run: |
-          chmod +x build_linux.sh
-          ./build_linux.sh ${{ inputs.flash-attn-version }} ${{ inputs.python-version }} ${{ inputs.torch-version }} ${{ inputs.cuda-version }}
-          wheel_path=$(ls flash-attention/dist/*.whl | head -n 1)
-          echo "WHEEL_PATH=$wheel_path" >> $GITHUB_OUTPUT
-
-      - name: Install Test
-        shell: bash
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload Release Asset
-        if: ${{ inputs.is-upload }}
-        shell: bash
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path="${{ steps.build_wheels.outputs.WHEEL_PATH }}"
-
-          # Check if the file exists
-          if [ ! -f "$wheel_path" ]; then
-            echo "Error: Wheel file not found at $wheel_path"
-            exit 1
-          fi
-
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path" --clobber
-
-      - name: Apply auditwheel repair
-        continue-on-error: true
-        id: auditwheel_repair
-        shell: bash
-        run: |
-          auditwheel show ${{ steps.build_wheels.outputs.WHEEL_PATH }}
-          auditwheel repair \
-            --exclude libc10* --exclude libtorch* --exclude libcu* --exclude libnv* --exclude 'libtorch*' \
-            ${{ steps.build_wheels.outputs.WHEEL_PATH }} -w flash-attention/dist_manylinux
-          ls -l flash-attention/dist_manylinux
-          wheel_path_manylinux=$(ls flash-attention/dist_manylinux/*manylinux*.whl | head -n 1)
-          echo "WHEEL_PATH_MANYLINUX=$wheel_path_manylinux" >> $GITHUB_OUTPUT
-
-      - name: Test manylinux wheel
-        continue-on-error: true
-        run: |
-          pip uninstall -y flash-attn > /dev/null 2>&1
-          pip install --no-cache-dir ${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}
-          python -c "import flash_attn; print(flash_attn.__version__)"
-
-      - name: Upload manylinux wheel
-        if: ${{ inputs.is-upload }}
-        continue-on-error: true
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          tag_name=${{ github.ref_name }}
-          wheel_path_manylinux="${{ steps.auditwheel_repair.outputs.WHEEL_PATH_MANYLINUX }}"
-          if [ ! -f "$wheel_path_manylinux" ]; then
-            echo "Error: Wheel file not found at $wheel_path_manylinux"
-            exit 1
-          fi
-          # Upload the release asset using GitHub CLI
-          gh release upload "$tag_name" "$wheel_path_manylinux" --clobber
-
-      - name: Clean up
-        shell: bash
-        if: always()
-        run: |
-          rm -rf /opt/hostedtoolcache/Python