ggml-python 0.0.38__tar.gz → 0.0.40__tar.gz

{ggml_python-0.0.38 → ggml_python-0.0.40}/.github/workflows/lint.yaml

@@ -12,10 +12,10 @@ jobs:
   ruff:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
 
       - name: Set up Python
-        uses: actions/setup-python@v5
+        uses: actions/setup-python@v6
         with:
           python-version: "3.12"
 

{ggml_python-0.0.38 → ggml_python-0.0.40}/.github/workflows/publish.yaml

@@ -2,7 +2,12 @@ name: Publish
 
 # Based on: https://packaging.python.org/en/latest/guides/publishing-package-distribution-releases-using-github-actions-ci-cd-workflows/
 
-on: workflow_dispatch
+on:
+  workflow_dispatch:
+  push:
+    tags:
+      - "v*.*.*"
+      - "!v*.*.*-*"
 
 jobs:
   build-n-publish:
@@ -10,11 +15,11 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-    - uses: actions/checkout@v4
+    - uses: actions/checkout@v6
       with:
         submodules: "true"
     - name: Set up Python
-      uses: actions/setup-python@v5
+      uses: actions/setup-python@v6
       with:
         python-version: "3.8"
     - name: Install dependencies
@@ -25,8 +30,6 @@ jobs:
       run: |
         python3 -m build --sdist
     - name: Publish distribution to PyPI
-      # TODO: move to tag based releases
-      # if: startsWith(github.ref, 'refs/tags')
       uses: pypa/gh-action-pypi-publish@release/v1
       with:
         password: ${{ secrets.PYPI_API_TOKEN }}

{ggml_python-0.0.38 → ggml_python-0.0.40}/.github/workflows/test.yaml

@@ -16,11 +16,11 @@ jobs:
         python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           submodules: "recursive"
       - name: Set up Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v5
+        uses: actions/setup-python@v6
         with:
           python-version: ${{ matrix.python-version }}
           cache: "pip"
@@ -38,6 +38,32 @@ jobs:
         run: |
           python -m pytest -s -vvvv
 
+  build-linux-onnx:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v6
+        with:
+          submodules: "recursive"
+      - name: Set up Python 3.11
+        uses: actions/setup-python@v6
+        with:
+          python-version: "3.11"
+          cache: "pip"
+      - name: Install dependencies
+        run: |
+          python3 -m pip install --upgrade pip pytest cmake scikit-build setuptools
+          python3 -m pip install \
+            --verbose \
+            --config-settings cmake.args='-DCMAKE_BUILD_TYPE=Debug;-DCMAKE_CXX_FLAGS=-g3;-DCMAKE_C_FLAGS=-g3' \
+            --config-settings cmake.verbose=true \
+            --config-settings logging.level=INFO \
+            --config-settings install.strip=false \
+            --editable ".[test,onnx,onnx-tests]"
+      - name: Test ONNX backend
+        run: |
+          python -m pytest -s -vvvv tests/test_ggml_onnx.py
+
   build-windows:
     runs-on: windows-latest
     strategy:
@@ -45,11 +71,11 @@ jobs:
         python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           submodules: "recursive"
       - name: Set up Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v5
+        uses: actions/setup-python@v6
         with:
           python-version: ${{ matrix.python-version }}
           cache: "pip"
@@ -69,11 +95,11 @@ jobs:
         python-version: ["3.9", "3.10", "3.11", "3.12"]
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           submodules: "recursive"
       - name: Set up Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v5
+        uses: actions/setup-python@v6
         with:
           python-version: ${{ matrix.python-version }}
           cache: "pip"
@@ -99,11 +125,11 @@ jobs:
   build-macos-intel:
     runs-on: macos-15-intel
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           submodules: "recursive"
       - name: Set up Python 3.9
-        uses: actions/setup-python@v5
+        uses: actions/setup-python@v6
         with:
           python-version: "3.9"
           cache: "pip"
@@ -130,11 +156,11 @@ jobs:
     runs-on: macos-15
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           submodules: "recursive"
       - name: Set up Python
-        uses: actions/setup-python@v5
+        uses: actions/setup-python@v6
         with:
           python-version: "3.9"
           cache: "pip"

{ggml_python-0.0.38 → ggml_python-0.0.40}/.github/workflows/wheels-cuda.yaml

@@ -1,6 +1,11 @@
 name: Wheels CUDA
 
-on: workflow_dispatch
+on:
+  workflow_dispatch:
+  push:
+    tags:
+      - "v*.*.*"
+      - "!v*.*.*-*"
 
 permissions:
   contents: write
@@ -63,17 +68,17 @@ jobs:
         with:
           arch: x64
 
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           submodules: "recursive"
 
-      - uses: actions/setup-python@v5
+      - uses: actions/setup-python@v6
         with:
           python-version: ${{ matrix.pyver }}
           cache: "pip"
 
       - name: Setup Mamba
-        uses: conda-incubator/setup-miniconda@v3.1.0
+        uses: conda-incubator/setup-miniconda@v4.0.1
         with:
           activate-environment: "ggml"
           python-version: ${{ matrix.pyver }}
@@ -224,7 +229,7 @@ jobs:
           # Publish tags that reflect the actual installed toolkit version.
           Write-Output "CUDA_VERSION=$cudaTagVersion" >> $env:GITHUB_ENV
 
-      - uses: softprops/action-gh-release@v2
+      - uses: softprops/action-gh-release@v3
         if: startsWith(github.ref, 'refs/tags/')
         with:
           files: dist/*

{ggml_python-0.0.38 → ggml_python-0.0.40}/.github/workflows/wheels-index.yaml

@@ -1,9 +1,11 @@
 name: Wheels Index
 
 on:
-  # Trigger on any new release
-  release:
-    types: [published]
+  # Trigger when release-asset workflows complete.
+  workflow_run:
+    workflows: ["Publish", "Wheels", "Wheels CUDA", "Wheels Metal", "Wheels Vulkan", "Wheels ROCm"]
+    types:
+      - completed
 
   # Allows you to run this workflow manually from the Actions tab
   workflow_dispatch:
@@ -29,11 +31,14 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
       - name: Setup Pages
-        uses: actions/configure-pages@v5
+        uses: actions/configure-pages@v6
       - name: Build
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
+          ./scripts/get-releases.sh
           ./scripts/releases-to-pep-503.sh index/whl/cpu '^[v]?[0-9]+\.[0-9]+\.[0-9]+$'
           ./scripts/releases-to-pep-503.sh index/whl/cu118 '^[v]?[0-9]+\.[0-9]+\.[0-9]+-cu118$'
           ./scripts/releases-to-pep-503.sh index/whl/cu121 '^[v]?[0-9]+\.[0-9]+\.[0-9]+-cu121$'
@@ -43,12 +48,15 @@ jobs:
           ./scripts/releases-to-pep-503.sh index/whl/cu125 '^[v]?[0-9]+\.[0-9]+\.[0-9]+-cu125$'
           ./scripts/releases-to-pep-503.sh index/whl/cu130 '^[v]?[0-9]+\.[0-9]+\.[0-9]+-cu130$'
           ./scripts/releases-to-pep-503.sh index/whl/cu132 '^[v]?[0-9]+\.[0-9]+\.[0-9]+-cu132$'
+          ./scripts/releases-to-pep-503.sh index/whl/vulkan '^[v]?[0-9]+\.[0-9]+\.[0-9]+-vulkan$'
+          ./scripts/releases-to-pep-503.sh index/whl/rocm72 '^[v]?[0-9]+\.[0-9]+\.[0-9]+-rocm72$'
+          ./scripts/releases-to-pep-503.sh index/whl/hip-radeon '^[v]?[0-9]+\.[0-9]+\.[0-9]+-hip-radeon$'
           ./scripts/releases-to-pep-503.sh index/whl/metal '^[v]?[0-9]+\.[0-9]+\.[0-9]+-metal$'
       - name: Upload artifact
-        uses: actions/upload-pages-artifact@v3
+        uses: actions/upload-pages-artifact@v5
         with:
           # Upload entire repository
-          path: "index"
+          path: 'index'
       - name: Deploy to GitHub Pages
         id: deployment
-        uses: actions/deploy-pages@v4
+        uses: actions/deploy-pages@v5

{ggml_python-0.0.38 → ggml_python-0.0.40}/.github/workflows/wheels-metal.yaml

@@ -1,6 +1,11 @@
 name: Wheels Metal
 
-on: workflow_dispatch
+on:
+  workflow_dispatch:
+  push:
+    tags:
+      - "v*.*.*"
+      - "!v*.*.*-*"
 
 permissions:
   contents: write
@@ -14,18 +19,18 @@ jobs:
         os: [macos-14, macos-15]
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           submodules: "recursive"
 
       # Used to host cibuildwheel.
-      - uses: actions/setup-python@v5
+      - uses: actions/setup-python@v6
         with:
           python-version: "3.12"
           cache: "pip"
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.22.0
+        uses: pypa/cibuildwheel@v3.4.1
         env:
           CIBW_REPAIR_WHEEL_COMMAND: ""
           CIBW_ARCHS: "arm64"
@@ -37,7 +42,7 @@ jobs:
           package-dir: .
           output-dir: wheelhouse
 
-      - uses: actions/upload-artifact@v4
+      - uses: actions/upload-artifact@v7
         with:
           name: wheels-mac_${{ matrix.os }}
           path: ./wheelhouse/*.whl
@@ -49,12 +54,12 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/download-artifact@v4
+      - uses: actions/download-artifact@v8
         with:
           merge-multiple: true
           path: dist
 
-      - uses: softprops/action-gh-release@v2
+      - uses: softprops/action-gh-release@v3
         with:
           files: dist/*
           # Set release name to <tag>-metal.

ggml_python-0.0.40/.github/workflows/wheels-rocm.yaml

@@ -0,0 +1,228 @@
+name: Wheels ROCm
+
+on:
+  workflow_dispatch:
+    inputs:
+      amdgpu_targets:
+        description: AMDGPU targets to compile into the Linux ROCm wheel
+        required: false
+        default: gfx908;gfx90a;gfx942;gfx1030;gfx1100;gfx1101;gfx1102;gfx1150;gfx1151;gfx1200;gfx1201
+      windows_amdgpu_targets:
+        description: AMDGPU targets to compile into the Windows HIP Radeon wheel
+        required: false
+        default: gfx1150;gfx1151;gfx1200;gfx1201;gfx1100;gfx1101;gfx1102;gfx1030;gfx1031;gfx1032
+  push:
+    tags:
+      - "v*.*.*"
+      - "!v*.*.*-*"
+
+permissions:
+  contents: write
+
+jobs:
+  build_wheels:
+    name: Build Wheel ${{ matrix.os }} ${{ matrix.pyver }} ROCm ${{ matrix.rocm }}
+    runs-on: ${{ matrix.os }}
+    container:
+      image: rocm/dev-ubuntu-22.04:${{ matrix.rocm }}-complete
+    strategy:
+      matrix:
+        include:
+          - os: ubuntu-22.04
+            pyver: "3.9"
+            rocm: "7.2.4"
+            amdgpu_targets: gfx908;gfx90a;gfx942;gfx1030;gfx1100;gfx1101;gfx1102;gfx1150;gfx1151;gfx1200;gfx1201
+
+    steps:
+      - name: Install system dependencies
+        run: |
+          apt-get update
+          apt-get install -y --no-install-recommends git cmake lsb-release ninja-build
+
+      - uses: actions/checkout@v6
+        with:
+          submodules: "recursive"
+
+      - uses: actions/setup-python@v6
+        with:
+          python-version: ${{ matrix.pyver }}
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install build wheel
+
+      - name: Build wheel
+        env:
+          MATRIX_AMDGPU_TARGETS: ${{ matrix.amdgpu_targets }}
+          INPUT_AMDGPU_TARGETS: ${{ inputs.amdgpu_targets }}
+        run: |
+          export ROCM_PATH="${ROCM_PATH:-/opt/rocm}"
+          export HIP_PATH="${HIP_PATH:-$ROCM_PATH}"
+          export PATH="$ROCM_PATH/bin:$ROCM_PATH/llvm/bin:$PATH"
+          export LD_LIBRARY_PATH="$ROCM_PATH/lib:$ROCM_PATH/lib64:${LD_LIBRARY_PATH:-}"
+          export CC="$ROCM_PATH/llvm/bin/clang"
+          export CXX="$ROCM_PATH/llvm/bin/clang++"
+          export HIPCXX="$ROCM_PATH/llvm/bin/clang"
+          export CMAKE_GENERATOR=Ninja
+
+          hipconfig --version
+          hipcc --version
+
+          rocm_tag="$(hipconfig --version | sed -E 's/^([0-9]+)\.([0-9]+).*/\1\2/')"
+          echo "ROCM_VERSION=$rocm_tag" >> "$GITHUB_ENV"
+
+          amdgpu_targets="${INPUT_AMDGPU_TARGETS:-$MATRIX_AMDGPU_TARGETS}"
+          export CMAKE_ARGS="-DGGML_HIP=on -DGGML_NATIVE=off -DGGML_AVX=off -DGGML_AVX2=off -DGGML_FMA=off -DGGML_F16C=off -DAMDGPU_TARGETS=$amdgpu_targets -DCMAKE_HIP_ARCHITECTURES=$amdgpu_targets"
+          python -m build --wheel
+
+      - uses: actions/upload-artifact@v7
+        with:
+          name: wheels-rocm${{ env.ROCM_VERSION }}-${{ matrix.os }}
+          path: dist/*.whl
+
+      - uses: softprops/action-gh-release@v3
+        if: startsWith(github.ref, 'refs/tags/')
+        with:
+          files: dist/*
+          # Set tag_name to <tag>-rocm<rocm_version>.
+          tag_name: ${{ github.ref_name }}-rocm${{ env.ROCM_VERSION }}
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+  build_wheels_windows_hip:
+    name: Build Wheel windows-2022 ${{ matrix.pyver }} HIP ${{ matrix.name }}
+    runs-on: windows-2022
+    env:
+      HIPSDK_INSTALLER_VERSION: "26.Q1"
+    strategy:
+      matrix:
+        include:
+          - name: radeon
+            pyver: "3.9"
+            amdgpu_targets: gfx1150;gfx1151;gfx1200;gfx1201;gfx1100;gfx1101;gfx1102;gfx1030;gfx1031;gfx1032
+
+    steps:
+      - uses: actions/checkout@v6
+        with:
+          submodules: "recursive"
+
+      - uses: actions/setup-python@v6
+        with:
+          python-version: ${{ matrix.pyver }}
+          cache: "pip"
+
+      - name: Grab rocWMMA package
+        run: |
+          curl -o rocwmma.deb "https://repo.radeon.com/rocm/apt/7.2.1/pool/main/r/rocwmma-dev/rocwmma-dev_2.2.0.70201-81~24.04_amd64.deb"
+          7z x rocwmma.deb
+          7z x data.tar
+
+      - name: Cache ROCm installation
+        id: cache-rocm
+        uses: actions/cache@v5
+        with:
+          path: C:\Program Files\AMD\ROCm
+          key: cache-gha-rocm-${{ env.HIPSDK_INSTALLER_VERSION }}-${{ runner.os }}
+
+      - name: Install ROCm
+        if: steps.cache-rocm.outputs.cache-hit != 'true'
+        run: |
+          $ErrorActionPreference = "Stop"
+          Write-Host "Downloading AMD HIP SDK installer"
+          Invoke-WebRequest `
+            -Uri "https://download.amd.com/developer/eula/rocm-hub/AMD-Software-PRO-Edition-${{ env.HIPSDK_INSTALLER_VERSION }}-Win11-For-HIP.exe" `
+            -OutFile "${env:RUNNER_TEMP}\rocm-install.exe"
+          Write-Host "Installing AMD HIP SDK"
+          $proc = Start-Process "${env:RUNNER_TEMP}\rocm-install.exe" -ArgumentList '-install' -NoNewWindow -PassThru
+          $completed = $proc.WaitForExit(1800000)
+          if (-not $completed) {
+            Write-Error "ROCm installation timed out after 30 minutes"
+            $proc.Kill()
+            exit 1
+          }
+          if ($proc.ExitCode -ne 0) {
+            Write-Error "ROCm installation failed with exit code $($proc.ExitCode)"
+            exit 1
+          }
+
+      - name: Verify ROCm
+        run: |
+          $clangPath = Get-ChildItem 'C:\Program Files\AMD\ROCm\*\bin\clang.exe' | Select-Object -First 1
+          if (-not $clangPath) {
+            Write-Error "ROCm installation not found"
+            exit 1
+          }
+          & $clangPath.FullName --version
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install build wheel
+
+      - name: Build wheel
+        env:
+          MATRIX_AMDGPU_TARGETS: ${{ matrix.amdgpu_targets }}
+          INPUT_AMDGPU_TARGETS: ${{ inputs.windows_amdgpu_targets }}
+        run: |
+          $ErrorActionPreference = "Stop"
+          $hipPath = Resolve-Path 'C:\Program Files\AMD\ROCm\*\bin\clang.exe' | Split-Path | Split-Path
+          $rocwmmaInclude = (Join-Path $PWD 'opt\rocm-7.2.1\include').Replace('\', '/')
+          $amdgpuTargets = if ($env:INPUT_AMDGPU_TARGETS) { $env:INPUT_AMDGPU_TARGETS } else { $env:MATRIX_AMDGPU_TARGETS }
+
+          $env:HIP_PATH = $hipPath
+          $env:ROCM_PATH = $hipPath
+          $env:CMAKE_PREFIX_PATH = $hipPath
+          $env:HIP_PLATFORM = 'amd'
+          $env:PATH = "$hipPath\bin;$env:PATH"
+          $env:CC = "$hipPath\bin\clang.exe"
+          $env:CXX = "$hipPath\bin\clang++.exe"
+          $env:HIPCXX = "$hipPath\bin\clang.exe"
+          $env:CMAKE_GENERATOR = 'Unix Makefiles'
+          $env:CXXFLAGS = "-I$rocwmmaInclude -Wno-ignored-attributes -Wno-nested-anon-types"
+          $env:CMAKE_ARGS = "-DGGML_HIP=ON -DGGML_HIP_ROCWMMA_FATTN=ON -DGGML_NATIVE=OFF -DGGML_AVX=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF -DGGML_F16C=OFF -DGPU_TARGETS=$amdgpuTargets"
+
+          python -m build --wheel
+
+      - name: Bundle ROCm runtime DLLs
+        run: |
+          $ErrorActionPreference = "Stop"
+          $hipPath = Resolve-Path 'C:\Program Files\AMD\ROCm\*\bin\clang.exe' | Split-Path | Split-Path
+          $wheel = Get-ChildItem dist\*.whl | Select-Object -First 1
+          python -m wheel unpack $wheel.FullName -d wheel-unpacked
+          $wheelRoot = Get-ChildItem wheel-unpacked -Directory | Select-Object -First 1
+          $libDir = Join-Path $wheelRoot.FullName 'ggml\lib'
+          New-Item -ItemType Directory -Force $libDir | Out-Null
+
+          $dllPatterns = @(
+            'amdhip64.dll',
+            'hiprtc*.dll',
+            'libhipblas.dll',
+            'libhipblaslt.dll',
+            'rocblas.dll'
+          )
+          foreach ($pattern in $dllPatterns) {
+            Copy-Item (Join-Path $hipPath "bin\$pattern") $libDir -ErrorAction SilentlyContinue
+          }
+
+          New-Item -ItemType Directory -Force (Join-Path $libDir 'rocblas\library') | Out-Null
+          New-Item -ItemType Directory -Force (Join-Path $libDir 'hipblaslt\library') | Out-Null
+          Copy-Item "$hipPath\bin\rocblas\library\*" (Join-Path $libDir 'rocblas\library') -Recurse -Force
+          Copy-Item "$hipPath\bin\hipblaslt\library\*" (Join-Path $libDir 'hipblaslt\library') -Recurse -Force
+
+          Remove-Item dist\*.whl
+          python -m wheel pack $wheelRoot.FullName -d dist
+
+      - uses: actions/upload-artifact@v7
+        with:
+          name: wheels-hip-${{ matrix.name }}-windows-2022
+          path: dist/*.whl
+
+      - uses: softprops/action-gh-release@v3
+        if: startsWith(github.ref, 'refs/tags/')
+        with:
+          files: dist/*
+          # Set tag_name to <tag>-hip-radeon.
+          tag_name: ${{ github.ref_name }}-hip-${{ matrix.name }}
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

ggml_python-0.0.40/.github/workflows/wheels-vulkan.yaml

@@ -0,0 +1,128 @@
+name: Wheels Vulkan
+
+on:
+  push:
+    tags:
+      - "v*.*.*"
+      - "!v*.*.*-*"
+  workflow_dispatch:
+
+permissions:
+  contents: write
+
+env:
+  VULKAN_SDK_VERSION: "1.4.341.0"
+  VULKAN_SDK_LINUX_SHA256: "ed66477d587a5587dc3601b1c2cdcc1fab5529c505f53a00171876cecd9b4fbe"
+
+jobs:
+  build_wheels:
+    name: Build Vulkan wheel on ${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - os: ubuntu-22.04
+            pyver: "3.9"
+            artifact: wheels-vulkan-ubuntu-22.04
+          - os: windows-2022
+            pyver: "3.9"
+            artifact: wheels-vulkan-windows-2022
+
+    steps:
+      - name: Set up MSVC
+        if: runner.os == 'Windows'
+        uses: ilammy/msvc-dev-cmd@v1
+        with:
+          arch: x64
+
+      - uses: actions/checkout@v6
+        with:
+          submodules: "recursive"
+
+      - uses: actions/setup-python@v6
+        with:
+          python-version: ${{ matrix.pyver }}
+          cache: "pip"
+
+      - name: Install Vulkan SDK
+        if: runner.os == 'Linux'
+        run: |
+          curl -fL \
+            "https://sdk.lunarg.com/sdk/download/${VULKAN_SDK_VERSION}/linux/vulkansdk-linux-x86_64-${VULKAN_SDK_VERSION}.tar.xz" \
+            -o vulkan-sdk.tar.xz
+          echo "${VULKAN_SDK_LINUX_SHA256}  vulkan-sdk.tar.xz" | sha256sum -c -
+          mkdir -p "$RUNNER_TEMP/vulkan-sdk"
+          tar -xf vulkan-sdk.tar.xz -C "$RUNNER_TEMP/vulkan-sdk"
+          source "$RUNNER_TEMP/vulkan-sdk/${VULKAN_SDK_VERSION}/setup-env.sh"
+          {
+            echo "VULKAN_SDK=$VULKAN_SDK"
+            echo "LD_LIBRARY_PATH=$VULKAN_SDK/lib:${LD_LIBRARY_PATH:-}"
+          } >> "$GITHUB_ENV"
+          echo "$VULKAN_SDK/bin" >> "$GITHUB_PATH"
+          "$VULKAN_SDK/bin/glslc" --version
+
+      - name: Install Vulkan SDK
+        if: runner.os == 'Windows'
+        shell: pwsh
+        run: |
+          choco install vulkan-sdk --version="$env:VULKAN_SDK_VERSION" --no-progress -y
+          $vulkanSdk = Join-Path 'C:\VulkanSDK' $env:VULKAN_SDK_VERSION
+          if (-not (Test-Path $vulkanSdk)) {
+            throw "Failed to find Vulkan SDK at $vulkanSdk"
+          }
+          "VULKAN_SDK=$vulkanSdk" >> $env:GITHUB_ENV
+          "$vulkanSdk\Bin" >> $env:GITHUB_PATH
+          & "$vulkanSdk\Bin\glslc.exe" --version
+
+      - name: Install build dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install build wheel
+
+      - name: Install Windows build dependencies
+        if: runner.os == 'Windows'
+        run: python -m pip install ninja
+
+      - name: Build Vulkan wheel
+        if: runner.os == 'Linux'
+        run: |
+          export CMAKE_ARGS="-DGGML_NATIVE=off -DGGML_METAL=OFF -DGGML_VULKAN=on"
+          python -m build --wheel
+          mkdir -p wheelhouse
+          cp dist/*.whl wheelhouse/
+
+      - name: Build Vulkan wheel
+        if: runner.os == 'Windows'
+        shell: pwsh
+        run: |
+          $env:CMAKE_GENERATOR = 'Ninja'
+          $env:CMAKE_ARGS = '-DGGML_NATIVE=off -DGGML_VULKAN=on'
+          python -m build --wheel
+          New-Item -ItemType Directory -Force wheelhouse | Out-Null
+          Copy-Item dist/*.whl wheelhouse/
+
+      - uses: actions/upload-artifact@v7
+        with:
+          name: ${{ matrix.artifact }}
+          path: ./wheelhouse/*.whl
+
+  release:
+    name: Release
+    needs: [build_wheels]
+    if: startsWith(github.ref, 'refs/tags/')
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/download-artifact@v8
+        with:
+          merge-multiple: true
+          path: dist
+
+      - uses: softprops/action-gh-release@v3
+        with:
+          files: dist/*
+          # Set release name to <tag>-vulkan.
+          tag_name: ${{ github.ref_name }}-vulkan
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}