feat: first commit

Signed-off-by: thxCode <[email protected]>
gpustack · Aug 11, 2024 · e359c61 · e359c61
commit e359c61
Show file tree

Hide file tree

Showing 92 changed files with 12,568 additions and 0 deletions.
diff --git a/.gitattributes b/.gitattributes
@@ -0,0 +1,4 @@
+* text=auto eol=lf
+
+**/go.sum linguist-generated=true
+**/zz_generated.*.go linguist-generated=true
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -0,0 +1,63 @@
+name: ci
+
+permissions:
+  contents: read
+  pull-requests: read
+  actions: read
+
+defaults:
+  run:
+    shell: bash
+
+on:
+  push:
+    branches:
+      - 'main'
+    paths-ignore:
+      - "docs/**"
+      - "**.md"
+      - "**.mdx"
+      - "**.png"
+      - "**.jpg"
+      - "packs/**"
+      - ".github/workflows/pack.yml"
+  pull_request:
+    branches:
+      - 'main'
+    paths-ignore:
+      - "docs/**"
+      - "**.md"
+      - "**.mdx"
+      - "**.png"
+      - "**.jpg"
+      - "packs/**"
+      - ".github/workflows/pack.yml"
+
+jobs:
+  ci:
+    timeout-minutes: 15
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+          persist-credentials: false
+      - name: Setup Go
+        timeout-minutes: 15
+        uses: actions/setup-go@v5
+        with:
+          go-version: "1.22.6"
+          cache-dependency-path: |
+            **/go.sum
+      - name: Setup Toolbox
+        timeout-minutes: 5
+        uses: actions/cache@v4
+        with:
+          key: toolbox-${{ runner.os }}
+          path: |
+            ${{ github.workspace }}/.sbin
+      - name: Make
+        run: make ci
+        env:
+          LINT_DIRTY: "true"
diff --git a/.github/workflows/cmd.yml b/.github/workflows/cmd.yml
@@ -0,0 +1,115 @@
+name: cmd
+
+permissions:
+  contents: write
+  actions: read
+  id-token: write
+
+defaults:
+  run:
+    shell: bash
+
+on:
+  push:
+    branches:
+      - main
+    paths-ignore:
+      - "docs/**"
+      - "**.md"
+      - "**.mdx"
+      - "**.png"
+      - "**.jpg"
+      - "packs/**"
+      - ".github/workflows/pack.yml"
+    tags:
+      - "v*.*.*"
+
+jobs:
+  build:
+    timeout-minutes: 15
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+          persist-credentials: false
+      - name: Setup Go
+        timeout-minutes: 15
+        uses: actions/setup-go@v5
+        with:
+          go-version: "1.22.6"
+          cache-dependency-path: |
+            **/go.sum
+      - name: Make
+        run: make build
+        env:
+          VERSION: "${{ github.ref_name }}"
+      - name: Upload Artifact
+        uses: actions/upload-artifact@v4
+        with:
+          path: ${{ github.workspace }}/.dist/*
+      - name: Release
+        if: ${{ startsWith(github.ref, 'refs/tags/') }}
+        uses: softprops/action-gh-release@v2
+        with:
+          fail_on_unmatched_files: true
+          tag_name: "${{ github.ref_name }}"
+          prerelease: ${{ contains(github.ref, 'rc') }}
+          files: ${{ github.workspace }}/.dist/*
+
+  publish:
+    needs:
+      - build
+    permissions:
+      contents: write
+      actions: read
+      id-token: write
+    timeout-minutes: 15
+    runs-on: ubuntu-22.04
+    env:
+      PACKAGE_REGISTRY: "thxcode"
+      PACKAGE_IMAGE: "gguf-packer"
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+          persist-credentials: false
+      - name: Setup QEMU
+        uses: docker/setup-qemu-action@v3
+        with:
+          image: tonistiigi/binfmt:qemu-v7.0.0
+          platforms: "arm64"
+      - name: Setup Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Login DockerHub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.CI_DOCKERHUB_USERNAME }}
+          password: ${{ secrets.CI_DOCKERHUB_PASSWORD }}
+      - name: Download Artifact
+        uses: actions/download-artifact@v4
+        with:
+          path: ${{ github.workspace }}/.dist
+          merge-multiple: true
+      - name: Get Metadata
+        id: metadata
+        uses: docker/metadata-action@v5
+        with:
+          images: "${{ env.PACKAGE_REGISTRY }}/${{ env.PACKAGE_IMAGE }}"
+      - name: Package
+        uses: docker/build-push-action@v6
+        with:
+          push: true
+          file: ${{ github.workspace }}/Dockerfile
+          context: ${{ github.workspace }}
+          platforms: "linux/amd64,linux/arm64"
+          tags: ${{ steps.metadata.outputs.tags }}
+          labels: ${{ steps.metadata.outputs.labels }}
+          cache-from: |
+            type=registry,ref=${{ env.PACKAGE_REGISTRY }}/${{ env.PACKAGE_IMAGE }}:build-cache
+          cache-to: |
+            type=registry,mode=max,oci-mediatypes=false,compression=gzip,ref=${{ env.PACKAGE_REGISTRY }}/${{ env.PACKAGE_IMAGE }}:build-cache,ignore-error=true
+          provenance: true
+          sbom: true
diff --git a/.github/workflows/pack.yml b/.github/workflows/pack.yml
@@ -0,0 +1,129 @@
+name: pack
+
+permissions:
+  contents: write
+  actions: read
+  id-token: write
+
+defaults:
+  run:
+    shell: bash
+
+on:
+  workflow_dispatch:
+    inputs:
+      model_repository:
+        description: "Model Repository"
+        required: true
+        type: string
+        default: "thxcode"
+      model_name:
+        description: "Model Name"
+        required: true
+        type: string
+        default: ""
+      model_tag:
+        description: "Model Tag"
+        required: true
+        type: string
+        default: ""
+      convert_huggingface_model:
+        description: "Convert HuggingFace Model"
+        type: string
+        default: ""
+
+jobs:
+  convert:
+    if: ${{ github.event.inputs.convert_huggingface_model != '' }}
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Maximize Space
+        # see https://github.com/easimon/maximize-build-space/blob/master/action.yml.
+        run: |
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf /usr/local/lib/android
+          sudo rm -rf /opt/ghc
+          sudo rm -rf /opt/hostedtoolcache/CodeQL
+          sudo docker image prune --all --force
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+          persist-credentials: false
+      - name: Download Model
+        env:
+          HF_TOKEN: ${{ secrets.CI_HUGGINGFACE_TOKEN }}
+        run: |
+          #!/bin/bash
+          git lfs install
+          git clone https://thxCode:${HF_TOKEN}@huggingface.co/${{ github.event.inputs.convert_huggingface_model }} ${{ github.workspace }}/packs/model
+      - name: Setup Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Login DockerHub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.CI_DOCKERHUB_USERNAME }}
+          password: ${{ secrets.CI_DOCKERHUB_PASSWORD }}
+      - name: Package
+        uses: docker/build-push-action@v6
+        with:
+          push: true
+          target: convert
+          file: "${{ github.workspace }}/packs/Dockerfile.${{ github.event.inputs.model_name }}"
+          context: "${{ github.workspace }}/packs"
+          tags: "${{ github.event.inputs.model_repository }}/${{ github.event.inputs.model_name }}:${{ github.event.inputs.model_tag }}-f16"
+          cache-from: |
+            type=registry,ref=${{ github.event.inputs.model_repository }}/build-cache:${{ github.event.inputs.model_name}}-${{ github.event.inputs.model_tag }}
+          cache-to: |
+            type=registry,ref=${{ github.event.inputs.model_repository }}/build-cache:${{ github.event.inputs.model_name}}-${{ github.event.inputs.model_tag }}
+
+  quantize:
+    needs:
+      - convert
+    strategy:
+      fail-fast: false
+      matrix:
+        type: [ Q3_K_M, Q4_K_M, Q5_K_M, Q6_K, Q8_0 ]
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Maximize Space
+        # see https://github.com/easimon/maximize-build-space/blob/master/action.yml.
+        run: |
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf /usr/local/lib/android
+          sudo rm -rf /opt/ghc
+          sudo rm -rf /opt/hostedtoolcache/CodeQL
+          sudo docker image prune --all --force
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+          persist-credentials: false
+      - name: Setup Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Login DockerHub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.CI_DOCKERHUB_USERNAME }}
+          password: ${{ secrets.CI_DOCKERHUB_PASSWORD }}
+      - name: Get Suffix
+        id: suffix
+        run: |
+          #!/bin/bash
+          suffix=$(echo "${QUANTIZE_TYPE}" | tr '[:upper:]' '[:lower:]' | sed 's/_/-/g')
+          echo "value=${suffix}" >> "$GITHUB_OUTPUT"
+      - name: Package
+        uses: docker/build-push-action@v6
+        with:
+          build-args: |
+            QUANTIZE_TYPE=${{ matrix.type }}
+          push: true
+          target: quantize
+          no-cache: true
+          file: "${{ github.workspace }}/packs/Dockerfile.${{ github.event.inputs.model_name }}"
+          context: "${{ github.workspace }}/packs"
+          tags: |
+            "${{ github.event.inputs.model_repository }}/${{ github.event.inputs.model_name }}:${{ github.event.inputs.model_tag }}-${{ steps.suffix.outputs.value }}"
+            "${{ matrix.type == 'Q4_K_M' && format('{0}/{1}:latest', github.event.inputs.model_repository, github.event.inputs.model_name) || '' }}"
+          cache-from: |
+            type=registry,ref=${{ github.event.inputs.model_repository }}/build-cache:${{ github.event.inputs.model_name}}-${{ github.event.inputs.model_tag }}
diff --git a/.gitignore b/.gitignore
@@ -0,0 +1,31 @@
+# Files
+.DS_Store
+*.lock
+*.test
+*.out
+*.swp
+*.swo
+*.db
+*.exe
+*.exe~
+*.dll
+*.so
+*.dylib
+*.log
+go.work
+go.work.*
+
+# Dirs
+/.idea
+/.vscode
+/.kube
+/.terraform
+/.vagrant
+/.bundle
+/.cache
+/.docker
+/.entc
+/.sbin
+/.dist
+/log
+/certs
diff --git a/.gitmodules b/.gitmodules
@@ -0,0 +1,15 @@
+[submodule "examples/models/Qwen2-0.5B-Instruct"]
+	path = examples/models/Qwen2-0.5B-Instruct
+	url = https://huggingface.co/Qwen/Qwen2-0.5B-Instruct
+[submodule "examples/ggufpackerfiles/single-stage-convert-from-local/Qwen2-0.5B-Instruct"]
+	path = examples/ggufpackerfiles/single-stage-convert-from-local/Qwen2-0.5B-Instruct
+	url = https://huggingface.co/Qwen/Qwen2-0.5B-Instruct
+[submodule "examples/ggufpackerfiles/multi-stages-convert-from-local/Qwen2-0.5B-Instruct"]
+	path = examples/ggufpackerfiles/multi-stages-convert-from-local/Qwen2-0.5B-Instruct
+	url = https://huggingface.co/Qwen/Qwen2-0.5B-Instruct
+[submodule "examples/ggufpackerfiles/multi-stages/Qwen2-0.5B-Instruct"]
+	path = examples/ggufpackerfiles/multi-stages/Qwen2-0.5B-Instruct
+	url = https://huggingface.co/Qwen/Qwen2-0.5B-Instruct
+[submodule "examples/ggufpackerfiles/single-stage/Qwen2-0.5B-Instruct"]
+	path = examples/ggufpackerfiles/single-stage/Qwen2-0.5B-Instruct
+	url = https://huggingface.co/Qwen/Qwen2-0.5B-Instruct