Revert "Revert #1963 (#2056)"

This reverts commit af9e5a2d05.

.dockerignore

@@ -1,4 +1,6 @@
 .idea
+.github
+.vscode
 models
 examples/chatbot-ui/models
 examples/rwkv/models

.env

@@ -1,33 +1,33 @@
 ## Set number of threads.
 ## Note: prefer the number of physical cores. Overbooking the CPU degrades performance notably.
-# THREADS=14
+# LOCALAI_THREADS=14
 
 ## Specify a different bind address (defaults to ":8080")
-# ADDRESS=127.0.0.1:8080
+# LOCALAI_ADDRESS=127.0.0.1:8080
 
 ## Default models context size
-# CONTEXT_SIZE=512
+# LOCALAI_CONTEXT_SIZE=512
 #
 ## Define galleries.
 ## models will to install will be visible in `/models/available`
-# GALLERIES=[{"name":"model-gallery", "url":"github:go-skynet/model-gallery/index.yaml"}]
+# LOCALAI_GALLERIES=[{"name":"model-gallery", "url":"github:go-skynet/model-gallery/index.yaml"}]
 
 ## CORS settings
-# CORS=true
-# CORS_ALLOW_ORIGINS=*
+# LOCALAI_CORS=true
+# LOCALAI_CORS_ALLOW_ORIGINS=*
 
 ## Default path for models
 #
-# MODELS_PATH=/models
+# LOCALAI_MODELS_PATH=/models
 
 ## Enable debug mode
-# DEBUG=true
+# LOCALAI_LOG_LEVEL=debug
 
 ## Disables COMPEL (Diffusers)
 # COMPEL=0
 
 ## Enable/Disable single backend (useful if only one GPU is available)
-# SINGLE_ACTIVE_BACKEND=true
+# LOCALAI_SINGLE_ACTIVE_BACKEND=true
 
 ## Specify a build type. Available: cublas, openblas, clblas.
 ## cuBLAS: This is a GPU-accelerated version of the complete standard BLAS (Basic Linear Algebra Subprograms) library. It's provided by Nvidia and is part of their CUDA toolkit.
@@ -46,13 +46,13 @@
 # GO_TAGS=stablediffusion
 
 ## Path where to store generated images
-# IMAGE_PATH=/tmp
+# LOCALAI_IMAGE_PATH=/tmp/generated/images
 
 ## Specify a default upload limit in MB (whisper)
-# UPLOAD_LIMIT
+# LOCALAI_UPLOAD_LIMIT=15
 
 ## List of external GRPC backends (note on the container image this variable is already set to use extra backends available in extra/)
-# EXTERNAL_GRPC_BACKENDS=my-backend:127.0.0.1:9000,my-backend2:/usr/bin/backend.py
+# LOCALAI_EXTERNAL_GRPC_BACKENDS=my-backend:127.0.0.1:9000,my-backend2:/usr/bin/backend.py
 
 ### Advanced settings ###
 ### Those are not really used by LocalAI, but from components in the stack ###
@@ -72,18 +72,18 @@
 # LLAMACPP_PARALLEL=1
 
 ### Enable to run parallel requests
-# PARALLEL_REQUESTS=true
+# LOCALAI_PARALLEL_REQUESTS=true
 
 ### Watchdog settings
 ###
 # Enables watchdog to kill backends that are inactive for too much time
-# WATCHDOG_IDLE=true
-#
-# Enables watchdog to kill backends that are busy for too much time
-# WATCHDOG_BUSY=true
+# LOCALAI_WATCHDOG_IDLE=true
 #
 # Time in duration format (e.g. 1h30m) after which a backend is considered idle
-# WATCHDOG_IDLE_TIMEOUT=5m
+# LOCALAI_WATCHDOG_IDLE_TIMEOUT=5m
+#
+# Enables watchdog to kill backends that are busy for too much time
+# LOCALAI_WATCHDOG_BUSY=true
 #
 # Time in duration format (e.g. 1h30m) after which a backend is considered busy
-# WATCHDOG_BUSY_TIMEOUT=5m
+# LOCALAI_WATCHDOG_BUSY_TIMEOUT=5m

.github/dependabot.yml

@@ -0,0 +1,25 @@
+# https://docs.github.com/en/code-security/dependabot/dependabot-version-updates/configuration-options-for-the-dependabot.yml-file
+version: 2
+updates:
+  - package-ecosystem: "gomod"
+    directory: "/"
+    schedule:
+      interval: "weekly"
+  - package-ecosystem: "github-actions"
+    # Workflow files stored in the default location of `.github/workflows`. (You don't need to specify `/.github/workflows` for `directory`. You can use `directory: "/"`.)
+    directory: "/"
+    schedule:
+      # Check for updates to GitHub Actions every weekday
+      interval: "weekly"
+  - package-ecosystem: "pip"
+    # Workflow files stored in the default location of `.github/workflows`. (You don't need to specify `/.github/workflows` for `directory`. You can use `directory: "/"`.)
+    directory: "/"
+    schedule:
+      # Check for updates to GitHub Actions every weekday
+      interval: "weekly"
+  - package-ecosystem: "docker"
+    # Workflow files stored in the default location of `.github/workflows`. (You don't need to specify `/.github/workflows` for `directory`. You can use `directory: "/"`.)
+    directory: "/"
+    schedule:
+      # Check for updates to GitHub Actions every weekday
+      interval: "weekly"

.github/labeler.yml

@@ -0,0 +1,19 @@
+enhancements:
+ - head-branch: ['^feature', 'feature']
+
+kind/documentation:
+- any:
+  - changed-files:
+    - any-glob-to-any-file: 'docs/*'
+  - changed-files:
+    - any-glob-to-any-file: '*.md'
+
+examples:
+- any:
+  - changed-files:
+    - any-glob-to-any-file: 'examples/*'
+
+ci:
+- any:
+  - changed-files:
+    - any-glob-to-any-file: '.github/*'

.github/workflows/bump_deps.yaml

@@ -49,7 +49,7 @@ jobs:
         run: |
           bash .github/bump_deps.sh ${{ matrix.repository }} ${{ matrix.branch }} ${{ matrix.variable }}
       - name: Create Pull Request
-        uses: peter-evans/create-pull-request@v5
+        uses: peter-evans/create-pull-request@v6
         with:
           token: ${{ secrets.UPDATE_BOT_TOKEN }}
           push-to-fork: ci-forks/LocalAI

.github/workflows/bump_docs.yaml

@@ -17,7 +17,7 @@ jobs:
         run: |
           bash .github/bump_docs.sh ${{ matrix.repository }}
       - name: Create Pull Request
-        uses: peter-evans/create-pull-request@v5
+        uses: peter-evans/create-pull-request@v6
         with:
           token: ${{ secrets.UPDATE_BOT_TOKEN }}
           push-to-fork: ci-forks/LocalAI

.github/workflows/dependabot_auto.yml

@@ -0,0 +1,43 @@
+name: Dependabot auto-merge
+on:
+- pull_request_target
+
+permissions:
+  contents: write
+  pull-requests: write
+  packages: read
+
+jobs:
+  dependabot:
+    runs-on: ubuntu-latest
+    if: ${{ github.actor == 'dependabot[bot]' }}
+    steps:
+      - name: Dependabot metadata
+        id: metadata
+        uses: dependabot/fetch-metadata@v2.0.0
+        with:
+          github-token: "${{ secrets.GITHUB_TOKEN }}"
+          skip-commit-verification: true
+
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Approve a PR if not already approved
+        run: |
+          gh pr checkout "$PR_URL"
+            if [ "$(gh pr status --json reviewDecision -q .currentBranch.reviewDecision)" != "APPROVED" ];
+          then
+            gh pr review --approve "$PR_URL"
+          else
+            echo "PR already approved.";
+          fi
+        env:
+          PR_URL: ${{github.event.pull_request.html_url}}
+          GITHUB_TOKEN: ${{secrets.GITHUB_TOKEN}}
+
+      - name: Enable auto-merge for Dependabot PRs
+        if: ${{ contains(github.event.pull_request.title, 'bump')}}
+        run: gh pr merge --auto --squash "$PR_URL"
+        env:
+          PR_URL: ${{github.event.pull_request.html_url}}
+          GITHUB_TOKEN: ${{secrets.GITHUB_TOKEN}}

.github/workflows/image-pr.yml

@@ -22,7 +22,7 @@ jobs:
       platforms: ${{ matrix.platforms }}
       runs-on: ${{ matrix.runs-on }}
       base-image: ${{ matrix.base-image }}
-      makeflags: "-j3"
+      makeflags: ${{ matrix.makeflags }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
       dockerPassword: ${{ secrets.DOCKERHUB_PASSWORD }}
@@ -42,6 +42,7 @@ jobs:
             image-type: 'extras'
             runs-on: 'arc-runner-set'
             base-image: "ubuntu:22.04"
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -52,6 +53,7 @@ jobs:
             image-type: 'extras'
             runs-on: 'arc-runner-set'
             base-image: "ubuntu:22.04"
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'hipblas'
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -60,6 +62,7 @@ jobs:
             image-type: 'extras'
             base-image: "rocm/dev-ubuntu-22.04:6.0-complete"
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'sycl_f16'
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -68,6 +71,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'extras'
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
   core-image-build:
     uses: ./.github/workflows/image_build.yml
     with:
@@ -81,7 +85,7 @@ jobs:
       platforms: ${{ matrix.platforms }}
       runs-on: ${{ matrix.runs-on }}
       base-image: ${{ matrix.base-image }}
-      makeflags: "-j3"
+      makeflags: ${{ matrix.makeflags }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
       dockerPassword: ${{ secrets.DOCKERHUB_PASSWORD }}
@@ -98,6 +102,7 @@ jobs:
             image-type: 'core'
             runs-on: 'ubuntu-latest'
             base-image: "ubuntu:22.04"
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'sycl_f16'
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -106,6 +111,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'core'
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -115,4 +121,5 @@ jobs:
             ffmpeg: 'true'
             image-type: 'core'
             runs-on: 'ubuntu-latest'
-            base-image: "ubuntu:22.04"
+            base-image: "ubuntu:22.04"
+            makeflags: "--jobs=5 --output-sync=target"

.github/workflows/image.yml

@@ -27,7 +27,9 @@ jobs:
       runs-on: ${{ matrix.runs-on }}
       base-image: ${{ matrix.base-image }}
       aio: ${{ matrix.aio }}
-      makeflags: "-j3"
+      makeflags: ${{ matrix.makeflags }}
+      latest-image: ${{ matrix.latest-image }}
+      latest-image-aio: ${{ matrix.latest-image-aio }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
       dockerPassword: ${{ secrets.DOCKERHUB_PASSWORD }}
@@ -49,6 +51,7 @@ jobs:
             image-type: 'extras'
             runs-on: 'arc-runner-set'
             base-image: "ubuntu:22.04"
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: ''
             platforms: 'linux/amd64'
             tag-latest: 'auto'
@@ -57,6 +60,7 @@ jobs:
             image-type: 'extras'
             runs-on: 'arc-runner-set'
             base-image: "ubuntu:22.04"
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "11"
             cuda-minor-version: "7"
@@ -67,6 +71,7 @@ jobs:
             image-type: 'extras'
             runs-on: 'arc-runner-set'
             base-image: "ubuntu:22.04"
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -77,6 +82,7 @@ jobs:
             image-type: 'extras'
             runs-on: 'arc-runner-set'
             base-image: "ubuntu:22.04"
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "11"
             cuda-minor-version: "7"
@@ -88,6 +94,9 @@ jobs:
             runs-on: 'arc-runner-set'
             base-image: "ubuntu:22.04"
             aio: "-aio-gpu-nvidia-cuda-11"
+            latest-image: 'latest-gpu-nvidia-cuda-11'
+            latest-image-aio: 'latest-aio-gpu-nvidia-cuda-11'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -99,6 +108,9 @@ jobs:
             runs-on: 'arc-runner-set'
             base-image: "ubuntu:22.04"
             aio: "-aio-gpu-nvidia-cuda-12"
+            latest-image: 'latest-gpu-nvidia-cuda-12'
+            latest-image-aio: 'latest-aio-gpu-nvidia-cuda-12'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: ''
             #platforms: 'linux/amd64,linux/arm64'
             platforms: 'linux/amd64'
@@ -108,6 +120,7 @@ jobs:
             image-type: 'extras'
             base-image: "ubuntu:22.04"
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'hipblas'
             platforms: 'linux/amd64'
             tag-latest: 'auto'
@@ -116,7 +129,10 @@ jobs:
             image-type: 'extras'
             aio: "-aio-gpu-hipblas"
             base-image: "rocm/dev-ubuntu-22.04:6.0-complete"
+            latest-image: 'latest-gpu-hipblas'
+            latest-image-aio: 'latest-aio-gpu-hipblas'
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'hipblas'
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -125,6 +141,7 @@ jobs:
             image-type: 'extras'
             base-image: "rocm/dev-ubuntu-22.04:6.0-complete"
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'sycl_f16'
             platforms: 'linux/amd64'
             tag-latest: 'auto'
@@ -134,6 +151,9 @@ jobs:
             image-type: 'extras'
             runs-on: 'arc-runner-set'
             aio: "-aio-gpu-intel-f16"
+            latest-image: 'latest-gpu-intel-f16'
+            latest-image-aio: 'latest-aio-gpu-intel-f16'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'sycl_f32'
             platforms: 'linux/amd64'
             tag-latest: 'auto'
@@ -143,6 +163,9 @@ jobs:
             image-type: 'extras'
             runs-on: 'arc-runner-set'
             aio: "-aio-gpu-intel-f32"
+            latest-image: 'latest-gpu-intel-f32'
+            latest-image-aio: 'latest-aio-gpu-intel-f32'
+            makeflags: "--jobs=3 --output-sync=target"
           # Core images
           - build-type: 'sycl_f16'
             platforms: 'linux/amd64'
@@ -152,6 +175,7 @@ jobs:
             ffmpeg: 'false'
             image-type: 'core'
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'sycl_f32'
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -160,6 +184,7 @@ jobs:
             ffmpeg: 'false'
             image-type: 'core'
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'sycl_f16'
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -168,6 +193,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'core'
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'sycl_f32'
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -176,6 +202,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'core'
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'hipblas'
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -184,6 +211,7 @@ jobs:
             image-type: 'core'
             base-image: "rocm/dev-ubuntu-22.04:6.0-complete"
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'hipblas'
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -192,6 +220,7 @@ jobs:
             image-type: 'core'
             base-image: "rocm/dev-ubuntu-22.04:6.0-complete"
             runs-on: 'arc-runner-set'
+            makeflags: "--jobs=3 --output-sync=target"
   
   core-image-build:
     uses: ./.github/workflows/image_build.yml
@@ -207,7 +236,9 @@ jobs:
       runs-on: ${{ matrix.runs-on }}
       aio: ${{ matrix.aio }}
       base-image: ${{ matrix.base-image }}
-      makeflags: "-j3"
+      makeflags: ${{ matrix.makeflags }}
+      latest-image: ${{ matrix.latest-image }}
+      latest-image-aio: ${{ matrix.latest-image-aio }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
       dockerPassword: ${{ secrets.DOCKERHUB_PASSWORD }}
@@ -225,6 +256,9 @@ jobs:
             base-image: "ubuntu:22.04"
             runs-on: 'ubuntu-latest'
             aio: "-aio-cpu"
+            latest-image: 'latest-cpu'
+            latest-image-aio: 'latest-aio-cpu'
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "11"
             cuda-minor-version: "7"
@@ -235,6 +269,7 @@ jobs:
             image-type: 'core'
             base-image: "ubuntu:22.04"
             runs-on: 'ubuntu-latest'
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -245,6 +280,7 @@ jobs:
             image-type: 'core'
             base-image: "ubuntu:22.04"
             runs-on: 'ubuntu-latest'
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "11"
             cuda-minor-version: "7"
@@ -255,6 +291,7 @@ jobs:
             image-type: 'core'
             runs-on: 'ubuntu-latest'
             base-image: "ubuntu:22.04"
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -265,3 +302,4 @@ jobs:
             image-type: 'core'
             runs-on: 'ubuntu-latest'
             base-image: "ubuntu:22.04"
+            makeflags: "--jobs=5 --output-sync=target"

.github/workflows/image_build.yml

@@ -29,6 +29,14 @@ on:
         description: 'Tag latest'
         default: ''
         type: string
+      latest-image:
+          description: 'Tag latest'
+          default: ''
+          type: string
+      latest-image-aio:
+          description: 'Tag latest'
+          default: ''
+          type: string
       tag-suffix:
         description: 'Tag suffix'
         default: ''
@@ -49,7 +57,7 @@ on:
       makeflags:
         description: 'Make Flags'
         required: false
-        default: ''
+        default: '--jobs=3 --output-sync=target'
         type: string
       aio:
         description: 'AIO Image Name'
@@ -79,6 +87,7 @@ jobs:
           && sudo apt-get install -y git
       - name: Checkout
         uses: actions/checkout@v4
+
       - name: Release space from worker
         if: inputs.runs-on == 'ubuntu-latest'
         run: |
@@ -120,6 +129,7 @@ jobs:
           sudo rm -rf "/usr/local/share/boost" || true
           sudo rm -rf "$AGENT_TOOLSDIRECTORY" || true
           df -h
+
       - name: Docker meta
         id: meta
         uses: docker/metadata-action@v5
@@ -134,6 +144,7 @@ jobs:
           flavor: |
             latest=${{ inputs.tag-latest }}
             suffix=${{ inputs.tag-suffix }}
+
       - name: Docker meta AIO (quay.io)
         if: inputs.aio != ''
         id: meta_aio
@@ -147,6 +158,7 @@ jobs:
           flavor: |
             latest=${{ inputs.tag-latest }}
             suffix=${{ inputs.aio }}
+
       - name: Docker meta AIO (dockerhub)
         if: inputs.aio != ''
         id: meta_aio_dockerhub
@@ -160,6 +172,7 @@ jobs:
           flavor: |
             latest=${{ inputs.tag-latest }}
             suffix=${{ inputs.aio }}
+
       - name: Set up QEMU
         uses: docker/setup-qemu-action@master
         with:
@@ -184,6 +197,25 @@ jobs:
           username: ${{ secrets.quayUsername }}
           password: ${{ secrets.quayPassword }}
 
+      - name: Cache GRPC
+        uses: docker/build-push-action@v5
+        with:
+          builder: ${{ steps.buildx.outputs.name }}
+          build-args: |
+            IMAGE_TYPE=${{ inputs.image-type }}
+            BASE_IMAGE=${{ inputs.base-image }}
+            MAKEFLAGS=${{ inputs.makeflags }}
+            GRPC_VERSION=v1.58.0
+          context: .
+          file: ./Dockerfile
+          cache-from: type=gha
+          cache-to: type=gha,ignore-error=true
+          target: grpc
+          platforms: ${{ inputs.platforms }}
+          push: false
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}
+
       - name: Build and push
         uses: docker/build-push-action@v5
         with:
@@ -198,18 +230,20 @@ jobs:
             MAKEFLAGS=${{ inputs.makeflags }}
           context: .
           file: ./Dockerfile
+          cache-from: type=gha
           platforms: ${{ inputs.platforms }}
           push: ${{ github.event_name != 'pull_request' }}
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}
-      -
-        name: Inspect image
+
+      - name: Inspect image
         if: github.event_name != 'pull_request'
         run: |
           docker pull localai/localai:${{ steps.meta.outputs.version }}
           docker image inspect localai/localai:${{ steps.meta.outputs.version }}
           docker pull quay.io/go-skynet/local-ai:${{ steps.meta.outputs.version }}
           docker image inspect quay.io/go-skynet/local-ai:${{ steps.meta.outputs.version }}
+
       - name: Build and push AIO image
         if: inputs.aio != ''
         uses: docker/build-push-action@v5
@@ -217,12 +251,14 @@ jobs:
           builder: ${{ steps.buildx.outputs.name }}
           build-args: |
             BASE_IMAGE=quay.io/go-skynet/local-ai:${{ steps.meta.outputs.version }}
+            MAKEFLAGS=${{ inputs.makeflags }}
           context: .
           file: ./Dockerfile.aio
           platforms: ${{ inputs.platforms }}
           push: ${{ github.event_name != 'pull_request' }}
           tags: ${{ steps.meta_aio.outputs.tags }}
           labels: ${{ steps.meta_aio.outputs.labels }}
+
       - name: Build and push AIO image (dockerhub)
         if: inputs.aio != ''
         uses: docker/build-push-action@v5
@@ -230,15 +266,39 @@ jobs:
           builder: ${{ steps.buildx.outputs.name }}
           build-args: |
             BASE_IMAGE=localai/localai:${{ steps.meta.outputs.version }}
+            MAKEFLAGS=${{ inputs.makeflags }}
           context: .
           file: ./Dockerfile.aio
           platforms: ${{ inputs.platforms }}
           push: ${{ github.event_name != 'pull_request' }}
           tags: ${{ steps.meta_aio_dockerhub.outputs.tags }}
           labels: ${{ steps.meta_aio_dockerhub.outputs.labels }}
+
+      - name: Latest tag
+        # run this on branches, when it is a tag and there is a latest-image defined
+        if: github.event_name != 'pull_request' && inputs.latest-image != ''  && github.ref_type == 'tag'
+        run: |
+          docker pull localai/localai:${{ steps.meta.outputs.version }}
+          docker tag localai/localai:${{ steps.meta.outputs.version }} localai/localai:${{ inputs.latest-image }}
+          docker push localai/localai:${{ inputs.latest-image }}
+          docker pull quay.io/go-skynet/local-ai:${{ steps.meta.outputs.version }}
+          docker tag quay.io/go-skynet/local-ai:${{ steps.meta.outputs.version }} quay.io/go-skynet/local-ai:${{ inputs.latest-image }}
+          docker push quay.io/go-skynet/local-ai:${{ inputs.latest-image }}
+      - name: Latest AIO tag
+        # run this on branches, when it is a tag and there is a latest-image defined
+        if: github.event_name != 'pull_request' && inputs.latest-image-aio != ''  && github.ref_type == 'tag'
+        run: |
+          docker pull localai/localai:${{ steps.meta_aio_dockerhub.outputs.version }}
+          docker tag localai/localai:${{ steps.meta_aio_dockerhub.outputs.version }} localai/localai:${{ inputs.latest-image-aio }}
+          docker push localai/localai:${{ inputs.latest-image-aio }}
+          docker pull quay.io/go-skynet/local-ai:${{ steps.meta_aio.outputs.version }}
+          docker tag quay.io/go-skynet/local-ai:${{ steps.meta_aio.outputs.version }} quay.io/go-skynet/local-ai:${{ inputs.latest-image-aio }}
+          docker push quay.io/go-skynet/local-ai:${{ inputs.latest-image-aio }}
+  
       - name: job summary
         run: |
           echo "Built image: ${{ steps.meta.outputs.labels }}" >> $GITHUB_STEP_SUMMARY
+
       - name: job summary(AIO)
         if: inputs.aio != ''
         run: |

.github/workflows/labeler.yml

@@ -0,0 +1,12 @@
+name: "Pull Request Labeler"
+on:
+- pull_request_target
+
+jobs:
+  labeler:
+    permissions:
+      contents: read
+      pull-requests: write
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/labeler@v5

.github/workflows/localaibot_automerge.yml

@@ -0,0 +1,35 @@
+name: LocalAI-bot auto-merge
+on:
+- pull_request_target
+
+permissions:
+  contents: write
+  pull-requests: write
+  packages: read
+
+jobs:
+  dependabot:
+    runs-on: ubuntu-latest
+    if: ${{ github.actor == 'localai-bot' }}
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Approve a PR if not already approved
+        run: |
+          gh pr checkout "$PR_URL"
+            if [ "$(gh pr status --json reviewDecision -q .currentBranch.reviewDecision)" != "APPROVED" ];
+          then
+            gh pr review --approve "$PR_URL"
+          else
+            echo "PR already approved.";
+          fi
+        env:
+          PR_URL: ${{github.event.pull_request.html_url}}
+          GITHUB_TOKEN: ${{secrets.GITHUB_TOKEN}}
+
+      - name: Enable auto-merge for LocalAIBot PRs
+        run: gh pr merge --auto --squash "$PR_URL"
+        env:
+          PR_URL: ${{github.event.pull_request.html_url}}
+          GITHUB_TOKEN: ${{secrets.GITHUB_TOKEN}}

.github/workflows/release.yaml

@@ -1,6 +1,11 @@
 name: Build and Release
 
-on: push
+on: 
+- push
+- pull_request
+
+env:
+  GRPC_VERSION: v1.58.0
 
 permissions:
   contents: write
@@ -30,13 +35,14 @@ jobs:
         uses: actions/checkout@v4
         with:
           submodules: true
-      - uses: actions/setup-go@v4
+      - uses: actions/setup-go@v5
         with:
-          go-version: '>=1.21.0'
+          go-version: '1.21.x'
+          cache: false
       - name: Dependencies
         run: |
           sudo apt-get update
-          sudo apt-get install build-essential ffmpeg
+          sudo apt-get install build-essential ffmpeg protobuf-compiler
       - name: Install CUDA Dependencies
         if: ${{ matrix.build == 'cuda12' || matrix.build == 'cuda11' }}
         run: |
@@ -51,26 +57,29 @@ jobs:
           sudo apt-get install -y cuda-nvcc-${CUDA_VERSION} libcublas-dev-${CUDA_VERSION}
       - name: Cache grpc
         id: cache-grpc
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: grpc
-          key: ${{ runner.os }}-grpc
+          key: ${{ runner.os }}-grpc-${{ env.GRPC_VERSION }}
       - name: Build grpc
         if: steps.cache-grpc.outputs.cache-hit != 'true'
         run: |
-          git clone --recurse-submodules -b v1.58.0 --depth 1 --shallow-submodules https://github.com/grpc/grpc && \
+          git clone --recurse-submodules -b ${{ env.GRPC_VERSION }} --depth 1 --shallow-submodules https://github.com/grpc/grpc && \
           cd grpc && mkdir -p cmake/build && cd cmake/build && cmake -DgRPC_INSTALL=ON \
             -DgRPC_BUILD_TESTS=OFF \
-            ../.. && sudo make -j12
+            ../.. && sudo make --jobs 5 --output-sync=target
       - name: Install gRPC
         run: |
-          cd grpc && cd cmake/build && sudo make -j12 install
+          cd grpc && cd cmake/build && sudo make --jobs 5 --output-sync=target install
       - name: Build
         id: build
         env:
           CMAKE_ARGS: "${{ matrix.defines }}"
           BUILD_ID: "${{ matrix.build }}"
         run: |
+          go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@latest
+          go install google.golang.org/protobuf/cmd/protoc-gen-go@latest
+          export PATH=$PATH:$GOPATH/bin
           if [ "${{ matrix.build }}" == "cuda12" ] || [ "${{ matrix.build }}" == "cuda11" ]; then
             export BUILD_TYPE=cublas
             export PATH=/usr/local/cuda/bin:$PATH
@@ -78,12 +87,12 @@ jobs:
           else
             STATIC=true make dist
           fi
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
-          name: ${{ matrix.build }}
+          name: LocalAI-linux-${{ matrix.build }}
           path: release/
       - name: Release
-        uses: softprops/action-gh-release@v1
+        uses: softprops/action-gh-release@v2
         if: startsWith(github.ref, 'refs/tags/')
         with:
           files: |
@@ -96,27 +105,24 @@ jobs:
         uses: actions/checkout@v4
         with:
           submodules: true
-      - uses: actions/setup-go@v4
+      - uses: actions/setup-go@v5
         with:
-          go-version: '>=1.21.0'
+          go-version: '1.21.x'
+          cache: false
       - name: Dependencies
         run: |
-          sudo apt-get install -y --no-install-recommends libopencv-dev
-          sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
+          sudo apt-get install -y --no-install-recommends libopencv-dev protobuf-compiler
+          go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@latest
+          go install google.golang.org/protobuf/cmd/protoc-gen-go@latest
       - name: Build stablediffusion
         run: |
+          export PATH=$PATH:$GOPATH/bin
           make backend-assets/grpc/stablediffusion
           mkdir -p release && cp backend-assets/grpc/stablediffusion release
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
           name: stablediffusion
           path: release/
-      - name: Release
-        uses: softprops/action-gh-release@v1
-        if: startsWith(github.ref, 'refs/tags/')
-        with:
-          files: |
-            release/*
 
   build-macOS:
     strategy:
@@ -134,12 +140,15 @@ jobs:
         uses: actions/checkout@v4
         with:
           submodules: true
-      - uses: actions/setup-go@v4
+      - uses: actions/setup-go@v5
         with:
-          go-version: '>=1.21.0'
+          go-version: '1.21.x'
+          cache: false
       - name: Dependencies
         run: |
           brew install protobuf grpc
+          go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@latest
+          go install google.golang.org/protobuf/cmd/protoc-gen-go@latest
       - name: Build
         id: build
         env:
@@ -148,13 +157,61 @@ jobs:
         run: |
           export C_INCLUDE_PATH=/usr/local/include
           export CPLUS_INCLUDE_PATH=/usr/local/include
+          export PATH=$PATH:$GOPATH/bin
           make dist
-      - uses: actions/upload-artifact@v3
+      - uses: actions/upload-artifact@v4
         with:
-          name: ${{ matrix.build }}
+          name: LocalAI-MacOS-${{ matrix.build }}
           path: release/
       - name: Release
-        uses: softprops/action-gh-release@v1
+        uses: softprops/action-gh-release@v2
+        if: startsWith(github.ref, 'refs/tags/')
+        with:
+          files: |
+            release/*
+
+
+  build-macOS-arm64:
+    strategy:
+      matrix:
+        include:
+          - build: 'avx2'
+            defines: ''
+          - build: 'avx'
+            defines: '-DLLAMA_AVX2=OFF'
+          - build: 'avx512'
+            defines: '-DLLAMA_AVX512=ON'
+    runs-on: macos-14
+    steps:
+      - name: Clone
+        uses: actions/checkout@v4
+        with:
+          submodules: true
+      - uses: actions/setup-go@v5
+        with:
+          go-version: '1.21.x'
+          cache: false
+      - name: Dependencies
+        run: |
+          brew install protobuf grpc
+          go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@latest
+          go install google.golang.org/protobuf/cmd/protoc-gen-go@latest
+      - name: Build
+        id: build
+        env:
+          CMAKE_ARGS: "${{ matrix.defines }}"
+          BUILD_ID: "${{ matrix.build }}"
+        run: |
+          export C_INCLUDE_PATH=/usr/local/include
+          export CPLUS_INCLUDE_PATH=/usr/local/include
+          export PATH=$PATH:$GOPATH/bin
+          make dist
+      - uses: actions/upload-artifact@v4
+        with:
+          name: LocalAI-MacOS-arm64-${{ matrix.build }}
+          path: release/
+      - name: Release
+        uses: softprops/action-gh-release@v2
         if: startsWith(github.ref, 'refs/tags/')
         with:
           files: |

.github/workflows/secscan.yaml

@@ -0,0 +1,30 @@
+name: "Security Scan"
+
+# Run workflow each time code is pushed to your repository and on a schedule.
+# The scheduled workflow runs every at 00:00 on Sunday UTC time.
+on:
+  push:
+  schedule:
+  - cron: '0 0 * * 0'
+
+jobs:
+  tests:
+    runs-on: ubuntu-latest
+    env:
+      GO111MODULE: on
+    steps:
+      - name: Checkout Source
+        uses: actions/checkout@v4
+        if: ${{ github.actor != 'dependabot[bot]' }}
+      - name: Run Gosec Security Scanner
+        if: ${{ github.actor != 'dependabot[bot]' }}
+        uses: securego/gosec@master
+        with:
+          # we let the report trigger content trigger a failure using the GitHub Security features.
+          args: '-no-fail -fmt sarif -out results.sarif ./...'
+      - name: Upload SARIF file
+        if: ${{ github.actor != 'dependabot[bot]' }}
+        uses: github/codeql-action/upload-sarif@v3
+        with:
+          # Path to SARIF file relative to the root of the repository
+          sarif_file: results.sarif

.github/workflows/test-extra.yml

@@ -32,16 +32,17 @@ jobs:
              sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
              sudo apt-get update && \
              sudo apt-get install -y conda
-          sudo apt-get install -y ca-certificates cmake curl patch
-          sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
+          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          sudo apt-get install -y libopencv-dev
+          pip install --user grpcio-tools
           
           sudo rm -rfv /usr/bin/conda || true
 
       - name: Test transformers
         run: |
            export PATH=$PATH:/opt/conda/bin
-           make -C backend/python/transformers
-           make -C backend/python/transformers test
+           make --jobs=5 --output-sync=target -C backend/python/transformers
+           make --jobs=5 --output-sync=target -C backend/python/transformers test
 
   tests-sentencetransformers:
     runs-on: ubuntu-latest
@@ -61,16 +62,17 @@ jobs:
              sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
              sudo apt-get update && \
              sudo apt-get install -y conda
-          sudo apt-get install -y ca-certificates cmake curl patch
-          sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
+          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          sudo apt-get install -y libopencv-dev
+          pip install --user grpcio-tools
           
           sudo rm -rfv /usr/bin/conda || true
 
       - name: Test sentencetransformers
         run: |
            export PATH=$PATH:/opt/conda/bin
-           make -C backend/python/sentencetransformers
-           make -C backend/python/sentencetransformers test
+           make --jobs=5 --output-sync=target -C backend/python/sentencetransformers
+           make --jobs=5 --output-sync=target -C backend/python/sentencetransformers test
 
   tests-diffusers:
     runs-on: ubuntu-latest
@@ -90,17 +92,47 @@ jobs:
              sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
              sudo apt-get update && \
              sudo apt-get install -y conda
-          sudo apt-get install -y ca-certificates cmake curl patch
-          sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
+          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          sudo apt-get install -y libopencv-dev
+          pip install --user grpcio-tools
           
           sudo rm -rfv /usr/bin/conda || true
 
       - name: Test diffusers
         run: |
            export PATH=$PATH:/opt/conda/bin
-           make -C backend/python/diffusers
-           make -C backend/python/diffusers test
+           make --jobs=5 --output-sync=target -C backend/python/diffusers
+           make --jobs=5 --output-sync=target -C backend/python/diffusers test
 
+  tests-parler-tts:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Clone
+        uses: actions/checkout@v4
+        with: 
+          submodules: true
+      - name: Dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install build-essential ffmpeg
+          curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
+             sudo install -o root -g root -m 644 conda.gpg /usr/share/keyrings/conda-archive-keyring.gpg && \
+              gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \
+             sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" > /etc/apt/sources.list.d/conda.list' && \
+             sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
+             sudo apt-get update && \
+             sudo apt-get install -y conda
+          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          sudo apt-get install -y libopencv-dev
+          pip install --user grpcio-tools
+          
+          sudo rm -rfv /usr/bin/conda || true
+
+      - name: Test parler-tts
+        run: |
+           export PATH=$PATH:/opt/conda/bin
+           make --jobs=5 --output-sync=target -C backend/python/parler-tts
+           make --jobs=5 --output-sync=target -C backend/python/parler-tts test
 
   tests-transformers-musicgen:
     runs-on: ubuntu-latest
@@ -120,47 +152,49 @@ jobs:
              sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
              sudo apt-get update && \
              sudo apt-get install -y conda
-          sudo apt-get install -y ca-certificates cmake curl patch
-          sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
+          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          sudo apt-get install -y libopencv-dev
+          pip install --user grpcio-tools
           
           sudo rm -rfv /usr/bin/conda || true
 
       - name: Test transformers-musicgen
         run: |
            export PATH=$PATH:/opt/conda/bin
-           make -C backend/python/transformers-musicgen
-           make -C backend/python/transformers-musicgen test
+           make --jobs=5 --output-sync=target -C backend/python/transformers-musicgen
+           make --jobs=5 --output-sync=target -C backend/python/transformers-musicgen test
 
 
 
-  tests-petals:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Clone
-        uses: actions/checkout@v4
-        with: 
-          submodules: true
-      - name: Dependencies
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential ffmpeg
-          curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
-             sudo install -o root -g root -m 644 conda.gpg /usr/share/keyrings/conda-archive-keyring.gpg && \
-              gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \
-             sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" > /etc/apt/sources.list.d/conda.list' && \
-             sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
-             sudo apt-get update && \
-             sudo apt-get install -y conda
-          sudo apt-get install -y ca-certificates cmake curl patch
-          sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
+  # tests-petals:
+  #   runs-on: ubuntu-latest
+  #   steps:
+  #     - name: Clone
+  #       uses: actions/checkout@v4
+  #       with: 
+  #         submodules: true
+  #     - name: Dependencies
+  #       run: |
+  #         sudo apt-get update
+  #         sudo apt-get install build-essential ffmpeg
+  #         curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
+  #            sudo install -o root -g root -m 644 conda.gpg /usr/share/keyrings/conda-archive-keyring.gpg && \
+  #             gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \
+  #            sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" > /etc/apt/sources.list.d/conda.list' && \
+  #            sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
+  #            sudo apt-get update && \
+  #            sudo apt-get install -y conda
+  #         sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+  #         sudo apt-get install -y libopencv-dev
+  #         pip install --user grpcio-tools
           
-          sudo rm -rfv /usr/bin/conda || true
+  #         sudo rm -rfv /usr/bin/conda || true
 
-      - name: Test petals
-        run: |
-           export PATH=$PATH:/opt/conda/bin
-           make -C backend/python/petals
-           make -C backend/python/petals test
+  #     - name: Test petals
+  #       run: |
+  #          export PATH=$PATH:/opt/conda/bin
+  #          make --jobs=5 --output-sync=target -C backend/python/petals
+  #          make --jobs=5 --output-sync=target -C backend/python/petals test
 
            
 
@@ -222,16 +256,17 @@ jobs:
   #            sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
   #            sudo apt-get update && \
   #            sudo apt-get install -y conda
-  #         sudo apt-get install -y ca-certificates cmake curl patch
-  #         sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
+  #         sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+  #         sudo apt-get install -y libopencv-dev
+  #         pip install --user grpcio-tools
           
   #         sudo rm -rfv /usr/bin/conda || true
 
   #     - name: Test bark
   #       run: |
   #          export PATH=$PATH:/opt/conda/bin
-  #          make -C backend/python/bark
-  #          make -C backend/python/bark test
+  #          make --jobs=5 --output-sync=target -C backend/python/bark
+  #          make --jobs=5 --output-sync=target -C backend/python/bark test
 
            
   # Below tests needs GPU. Commented out for now
@@ -254,14 +289,15 @@ jobs:
   #            sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
   #            sudo apt-get update && \
   #            sudo apt-get install -y conda
-  #         sudo apt-get install -y ca-certificates cmake curl patch
-  #         sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
+  #         sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+  #         sudo apt-get install -y libopencv-dev
+  #         pip install --user grpcio-tools
   #         sudo rm -rfv /usr/bin/conda || true
   #     - name: Test vllm
   #       run: |
   #          export PATH=$PATH:/opt/conda/bin
-  #          make -C backend/python/vllm
-  #          make -C backend/python/vllm test
+  #          make --jobs=5 --output-sync=target -C backend/python/vllm
+  #          make --jobs=5 --output-sync=target -C backend/python/vllm test
   tests-vallex:
     runs-on: ubuntu-latest
     steps:
@@ -280,14 +316,15 @@ jobs:
              sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
              sudo apt-get update && \
              sudo apt-get install -y conda
-          sudo apt-get install -y ca-certificates cmake curl patch
-          sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2    
+          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          sudo apt-get install -y libopencv-dev
+          pip install --user grpcio-tools
           sudo rm -rfv /usr/bin/conda || true
       - name: Test vall-e-x
         run: |
            export PATH=$PATH:/opt/conda/bin
-           make -C backend/python/vall-e-x
-           make -C backend/python/vall-e-x test
+           make --jobs=5 --output-sync=target -C backend/python/vall-e-x
+           make --jobs=5 --output-sync=target -C backend/python/vall-e-x test
 
   tests-coqui:
     runs-on: ubuntu-latest
@@ -307,11 +344,12 @@ jobs:
              sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
              sudo apt-get update && \
              sudo apt-get install -y conda
-          sudo apt-get install -y ca-certificates cmake curl patch espeak espeak-ng          
+          sudo apt-get install -y ca-certificates cmake curl patch espeak espeak-ng python3-pip
+          pip install --user grpcio-tools
           sudo rm -rfv /usr/bin/conda || true
 
       - name: Test coqui
         run: |
            export PATH=$PATH:/opt/conda/bin
-           make -C backend/python/coqui
-           make -C backend/python/coqui test
+           make --jobs=5 --output-sync=target -C backend/python/coqui
+           make --jobs=5 --output-sync=target -C backend/python/coqui test

.github/workflows/test.yml

@@ -9,6 +9,9 @@ on:
     tags:
       - '*'
 
+env:
+  GRPC_VERSION: v1.58.0
+
 concurrency:
   group: ci-tests-${{ github.head_ref || github.ref }}-${{ github.repository }}
   cancel-in-progress: true
@@ -57,26 +60,37 @@ jobs:
         with: 
           submodules: true
       - name: Setup Go ${{ matrix.go-version }}
-        uses: actions/setup-go@v4
+        uses: actions/setup-go@v5
         with:
           go-version: ${{ matrix.go-version }}
+          cache: false
       # You can test your matrix by printing the current Go version
       - name: Display Go version
         run: go version
       - name: Dependencies
         run: |
           sudo apt-get update
-          sudo apt-get install build-essential ffmpeg
+          sudo apt-get install build-essential curl ffmpeg
           curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
              sudo install -o root -g root -m 644 conda.gpg /usr/share/keyrings/conda-archive-keyring.gpg && \
-              gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \
+             gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \
              sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" > /etc/apt/sources.list.d/conda.list' && \
              sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
              sudo apt-get update && \
              sudo apt-get install -y conda
-          sudo apt-get install -y ca-certificates cmake curl patch
-          sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
-          
+          sudo apt-get install -y ca-certificates cmake patch python3-pip unzip
+          sudo apt-get install -y libopencv-dev
+
+          curl -L -s https://github.com/protocolbuffers/protobuf/releases/download/v26.1/protoc-26.1-linux-x86_64.zip -o protoc.zip && \
+          unzip -j -d /usr/local/bin protoc.zip bin/protoc && \
+          rm protoc.zip
+
+          go install google.golang.org/protobuf/cmd/protoc-gen-go@latest
+          go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@latest
+
+          # The python3-grpc-tools package in 22.04 is too old
+          pip install --user grpcio-tools
+
           sudo rm -rfv /usr/bin/conda || true
           PATH=$PATH:/opt/conda/bin make -C backend/python/sentencetransformers
 
@@ -85,30 +99,31 @@ jobs:
           GO_TAGS="tts" make -C sources/go-piper piper.o && \
           sudo cp -rfv sources/go-piper/piper-phonemize/pi/lib/. /usr/lib/ && \
           # Pre-build stable diffusion before we install a newer version of abseil (not compatible with stablediffusion-ncn)
-          GO_TAGS="stablediffusion tts" GRPC_BACKENDS=backend-assets/grpc/stablediffusion make build
+          PATH="$PATH:/root/go/bin" GO_TAGS="stablediffusion tts" GRPC_BACKENDS=backend-assets/grpc/stablediffusion make build
       - name: Cache grpc
         id: cache-grpc
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: grpc
-          key: ${{ runner.os }}-grpc
+          key: ${{ runner.os }}-grpc-${{ env.GRPC_VERSION }}
       - name: Build grpc
         if: steps.cache-grpc.outputs.cache-hit != 'true'
         run: |
-          git clone --recurse-submodules -b v1.58.0 --depth 1 --shallow-submodules https://github.com/grpc/grpc && \
+          git clone --recurse-submodules -b ${{ env.GRPC_VERSION }} --depth 1 --jobs 5 --shallow-submodules https://github.com/grpc/grpc && \
           cd grpc && mkdir -p cmake/build && cd cmake/build && cmake -DgRPC_INSTALL=ON \
             -DgRPC_BUILD_TESTS=OFF \
-            ../.. && sudo make -j12
+            ../.. && sudo make --jobs 5
       - name: Install gRPC
         run: |
-          cd grpc && cd cmake/build && sudo make -j12 install
+          cd grpc && cd cmake/build && sudo make --jobs 5 install
       - name: Test
         run: |
-          GO_TAGS="stablediffusion tts" make test
+          PATH="$PATH:/root/go/bin" GO_TAGS="stablediffusion tts" make --jobs 5 --output-sync=target test
       - name: Setup tmate session if tests fail
         if: ${{ failure() }}
-        uses: mxschmitt/action-tmate@v3
-        timeout-minutes: 5
+        uses: mxschmitt/action-tmate@v3.18
+        with:
+          connect-timeout-seconds: 180
 
   tests-aio-container:
     runs-on: ubuntu-latest
@@ -151,7 +166,7 @@ jobs:
           submodules: true
       - name: Build images
         run: |
-          docker build --build-arg FFMPEG=true --build-arg IMAGE_TYPE=core -t local-ai:tests -f Dockerfile .
+          docker build --build-arg FFMPEG=true --build-arg IMAGE_TYPE=core --build-arg MAKEFLAGS="--jobs=5 --output-sync=target" -t local-ai:tests -f Dockerfile .
           BASE_IMAGE=local-ai:tests DOCKER_AIO_IMAGE=local-ai-aio:test make docker-aio
       - name: Test
         run: |
@@ -159,8 +174,9 @@ jobs:
             make run-e2e-aio
       - name: Setup tmate session if tests fail
         if: ${{ failure() }}
-        uses: mxschmitt/action-tmate@v3
-        timeout-minutes: 5
+        uses: mxschmitt/action-tmate@v3.18
+        with:
+          connect-timeout-seconds: 180
 
   tests-apple:
     runs-on: macOS-14
@@ -173,21 +189,26 @@ jobs:
         with: 
           submodules: true
       - name: Setup Go ${{ matrix.go-version }}
-        uses: actions/setup-go@v4
+        uses: actions/setup-go@v5
         with:
           go-version: ${{ matrix.go-version }}
+          cache: false
       # You can test your matrix by printing the current Go version
       - name: Display Go version
         run: go version
       - name: Dependencies
         run: |
-          brew install protobuf grpc
+          brew install protobuf grpc make protoc-gen-go protoc-gen-go-grpc
+          pip install --user grpcio-tools
       - name: Test
         run: |
           export C_INCLUDE_PATH=/usr/local/include
           export CPLUS_INCLUDE_PATH=/usr/local/include
-          BUILD_TYPE="GITHUB_CI_HAS_BROKEN_METAL" CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF" make test
+          # Used to run the newer GNUMake version from brew that supports --output-sync
+          export PATH="/opt/homebrew/opt/make/libexec/gnubin:$PATH"
+          BUILD_TYPE="GITHUB_CI_HAS_BROKEN_METAL" CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF" make --jobs 4 --output-sync=target test
       - name: Setup tmate session if tests fail
         if: ${{ failure() }}
-        uses: mxschmitt/action-tmate@v3
-        timeout-minutes: 5
+        uses: mxschmitt/action-tmate@v3.18
+        with:
+          connect-timeout-seconds: 180

.gitignore

@@ -39,3 +39,8 @@ backend-assets/*
 !backend-assets/.keep
 prepare
 /ggml-metal.metal
+
+# Protobuf generated files
+*.pb.go
+*pb2.py
+*pb2_grpc.py

.vscode/extensions.json

@@ -0,0 +1,5 @@
+{
+    "recommendations": [
+        "golang.go"
+    ]
+}

CONTRIBUTING.md

@@ -1,4 +1,4 @@
-# Contributing to localAI
+# Contributing to LocalAI
 
 Thank you for your interest in contributing to LocalAI! We appreciate your time and effort in helping to improve our project. Before you get started, please take a moment to review these guidelines.
 
@@ -29,8 +29,9 @@ Thank you for your interest in contributing to LocalAI! We appreciate your time
 
 1. Clone the repository: `git clone https://github.com/go-skynet/LocalAI.git`
 2. Navigate to the project directory: `cd LocalAI`
-3. Install the required dependencies: `make prepare`
-4. Run LocalAI: `make run`
+3. Install the required dependencies ( see https://localai.io/basics/build/#build-localai-locally )
+4. Build LocalAI: `make build`
+5. Run LocalAI: `./local-ai`
 
 ## Contributing
 
@@ -59,14 +60,29 @@ If you find a bug, have a feature request, or encounter any issues, please check
 
 `make test` cannot handle all the model now. Please be sure to add a test case for the new features or the part was changed.
 
+### Running AIO tests
+
+All-In-One images has a set of tests that automatically verifies that most of the endpoints works correctly, a flow can be :
+
+```bash
+# Build the LocalAI docker image
+make DOCKER_IMAGE=local-ai docker
+
+# Build the corresponding AIO image
+BASE_IMAGE=local-ai DOCKER_AIO_IMAGE=local-ai-aio:test make docker-aio
+
+# Run the AIO e2e tests
+LOCALAI_IMAGE_TAG=test LOCALAI_IMAGE=local-ai-aio make run-e2e-aio
+```
+
 ## Documentation
 
-- We are welcome the contribution of the documents, please open new PR in the official document repo [localai-website](https://github.com/go-skynet/localai-website)
-
+We are welcome the contribution of the documents, please open new PR or create a new issue. The documentation is available under `docs/` https://github.com/mudler/LocalAI/tree/master/docs
+ 
 ## Community and Communication
 
 - You can reach out via the Github issue tracker.
 - Open a new discussion at [Discussion](https://github.com/go-skynet/LocalAI/discussions)
 - Join the Discord channel [Discord](https://discord.gg/uJAeKSAGDy)
 
----
+---

Dockerfile

@@ -15,17 +15,30 @@ ARG TARGETVARIANT
 
 ENV BUILD_TYPE=${BUILD_TYPE}
 ENV DEBIAN_FRONTEND=noninteractive
-ENV EXTERNAL_GRPC_BACKENDS="coqui:/build/backend/python/coqui/run.sh,huggingface-embeddings:/build/backend/python/sentencetransformers/run.sh,petals:/build/backend/python/petals/run.sh,transformers:/build/backend/python/transformers/run.sh,sentencetransformers:/build/backend/python/sentencetransformers/run.sh,autogptq:/build/backend/python/autogptq/run.sh,bark:/build/backend/python/bark/run.sh,diffusers:/build/backend/python/diffusers/run.sh,exllama:/build/backend/python/exllama/run.sh,vall-e-x:/build/backend/python/vall-e-x/run.sh,vllm:/build/backend/python/vllm/run.sh,mamba:/build/backend/python/mamba/run.sh,exllama2:/build/backend/python/exllama2/run.sh,transformers-musicgen:/build/backend/python/transformers-musicgen/run.sh"
+ENV EXTERNAL_GRPC_BACKENDS="coqui:/build/backend/python/coqui/run.sh,huggingface-embeddings:/build/backend/python/sentencetransformers/run.sh,petals:/build/backend/python/petals/run.sh,transformers:/build/backend/python/transformers/run.sh,sentencetransformers:/build/backend/python/sentencetransformers/run.sh,autogptq:/build/backend/python/autogptq/run.sh,bark:/build/backend/python/bark/run.sh,diffusers:/build/backend/python/diffusers/run.sh,exllama:/build/backend/python/exllama/run.sh,vall-e-x:/build/backend/python/vall-e-x/run.sh,vllm:/build/backend/python/vllm/run.sh,mamba:/build/backend/python/mamba/run.sh,exllama2:/build/backend/python/exllama2/run.sh,transformers-musicgen:/build/backend/python/transformers-musicgen/run.sh,parler-tts:/build/backend/python/parler-tts/run.sh"
 
 ARG GO_TAGS="stablediffusion tinydream tts"
 
 RUN apt-get update && \
-    apt-get install -y ca-certificates curl patch pip cmake git && apt-get clean
+    apt-get install -y ca-certificates curl python3-pip unzip && apt-get clean
 
 # Install Go
 RUN curl -L -s https://go.dev/dl/go$GO_VERSION.linux-$TARGETARCH.tar.gz | tar -C /usr/local -xz
 ENV PATH $PATH:/usr/local/go/bin
 
+# Install grpc compilers
+ENV PATH $PATH:/root/go/bin
+RUN go install google.golang.org/protobuf/cmd/protoc-gen-go@latest && \
+    go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@latest
+
+# Install protobuf (the version in 22.04 is too old)
+RUN curl -L -s https://github.com/protocolbuffers/protobuf/releases/download/v26.1/protoc-26.1-linux-x86_64.zip -o protoc.zip && \
+    unzip -j -d /usr/local/bin protoc.zip bin/protoc && \
+    rm protoc.zip
+
+# Install grpcio-tools (the version in 22.04 is too old)
+RUN pip install --user grpcio-tools
+
 COPY --chmod=644 custom-ca-certs/* /usr/local/share/ca-certificates/
 RUN update-ca-certificates
 
@@ -68,7 +81,8 @@ RUN test -n "$TARGETARCH" \
 
 FROM requirements-core as requirements-extras
 
-RUN curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
+RUN apt install -y gpg && \
+    curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
     install -o root -g root -m 644 conda.gpg /usr/share/keyrings/conda-archive-keyring.gpg && \
     gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \
     echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" > /etc/apt/sources.list.d/conda.list && \
@@ -90,11 +104,35 @@ RUN if [ ! -e /usr/bin/python ]; then \
 ###################################
 ###################################
 
+FROM ${BASE_IMAGE} as grpc
+
+ARG MAKEFLAGS
+ARG GRPC_VERSION=v1.58.0
+
+ENV MAKEFLAGS=${MAKEFLAGS}
+
+WORKDIR /build
+
+RUN apt-get update && \
+    apt-get install -y build-essential cmake git  && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/*
+
+RUN git clone --recurse-submodules --jobs 4 -b ${GRPC_VERSION} --depth 1 --shallow-submodules https://github.com/grpc/grpc
+
+RUN cd grpc && \
+    mkdir -p cmake/build && \
+    cd cmake/build && \
+    cmake -DgRPC_INSTALL=ON -DgRPC_BUILD_TESTS=OFF ../.. && \
+    make
+
+###################################
+###################################
+
 FROM requirements-${IMAGE_TYPE} as builder
 
 ARG GO_TAGS="stablediffusion tts"
 ARG GRPC_BACKENDS
-ARG BUILD_GRPC=true
 ARG MAKEFLAGS
 
 ENV GRPC_BACKENDS=${GRPC_BACKENDS}
@@ -109,6 +147,12 @@ WORKDIR /build
 COPY . .
 COPY .git .
 RUN echo "GO_TAGS: $GO_TAGS"
+
+RUN apt-get update && \
+    apt-get install -y build-essential cmake git  && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/*
+
 RUN make prepare
 
 # If we are building with clblas support, we need the libraries for the builds
@@ -121,12 +165,9 @@ RUN if [ "${BUILD_TYPE}" = "clblas" ]; then \
 # stablediffusion does not tolerate a newer version of abseil, build it first
 RUN GRPC_BACKENDS=backend-assets/grpc/stablediffusion make build
 
-RUN if [ "${BUILD_GRPC}" = "true" ]; then \
-    git clone --recurse-submodules --jobs 4 -b v1.58.0 --depth 1 --shallow-submodules https://github.com/grpc/grpc && \
-    cd grpc && mkdir -p cmake/build && cd cmake/build && cmake -DgRPC_INSTALL=ON \
-      -DgRPC_BUILD_TESTS=OFF \
-       ../.. && make install \
-    ; fi
+COPY --from=grpc /build/grpc ./grpc/
+
+RUN cd /build/grpc/cmake/build && make install
 
 # Rebuild with defaults backends
 RUN make build
@@ -170,6 +211,11 @@ RUN if [ "${BUILD_TYPE}" = "clblas" ]; then \
     apt-get clean \
     ; fi
 
+RUN apt-get update && \
+    apt-get install -y cmake git  && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/*
+
 WORKDIR /build
 
 # we start fresh & re-copy all assets because `make build` does not clean up nicely after itself
@@ -179,9 +225,9 @@ WORKDIR /build
 COPY . .
 
 COPY --from=builder /build/sources ./sources/
-COPY --from=builder /build/grpc ./grpc/
+COPY --from=grpc /build/grpc ./grpc/
 
-RUN make prepare-sources && cd /build/grpc/cmake/build && make install && rm -rf grpc
+RUN make prepare-sources && cd /build/grpc/cmake/build && make install && rm -rf /build/grpc
 
 # Copy the binary
 COPY --from=builder /build/local-ai ./
@@ -229,6 +275,9 @@ RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
 RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
     make -C backend/python/transformers-musicgen \
     ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/parler-tts \
+    ; fi
 RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
     make -C backend/python/coqui \
     ; fi

Makefile

@@ -5,7 +5,7 @@ BINARY_NAME=local-ai
 
 # llama.cpp versions
 GOLLAMA_STABLE_VERSION?=2b57a8ae43e4699d3dc5d1496a1ccd42922993be
-CPPLLAMA_VERSION?=b06c16ef9f81d84da520232c125d4d8a1d273736
+CPPLLAMA_VERSION?=7593639ce335e8d7f89aa9a54d616951f273af60
 
 # gpt4all version
 GPT4ALL_REPO?=https://github.com/nomic-ai/gpt4all
@@ -16,7 +16,7 @@ RWKV_REPO?=https://github.com/donomii/go-rwkv.cpp
 RWKV_VERSION?=661e7ae26d442f5cfebd2a0881b44e8c55949ec6
 
 # whisper.cpp version
-WHISPER_CPP_VERSION?=1558ec5a16cb2b2a0bf54815df1d41f83dc3815b
+WHISPER_CPP_VERSION?=b0c3cbf2e851cf232e432b590dcc514a689ec028
 
 # bert.cpp version
 BERT_VERSION?=6abe312cded14042f6b7c3cd8edf082713334a4d
@@ -28,7 +28,7 @@ PIPER_VERSION?=9d0100873a7dbb0824dfea40e8cec70a1b110759
 STABLEDIFFUSION_VERSION?=362df9da29f882dbf09ade61972d16a1f53c3485
 
 # tinydream version
-TINYDREAM_VERSION?=772a9c0d9aaf768290e63cca3c904fe69faf677a
+TINYDREAM_VERSION?=22a12a4bc0ac5455856f28f3b771331a551a4293
 
 export BUILD_TYPE?=
 export STABLE_BUILD_TYPE?=$(BUILD_TYPE)
@@ -224,7 +224,7 @@ sources/go-stable-diffusion:
 	cd sources/go-stable-diffusion && git checkout -b build $(STABLEDIFFUSION_VERSION) && git submodule update --init --recursive --depth 1
 
 sources/go-stable-diffusion/libstablediffusion.a: sources/go-stable-diffusion
-	$(MAKE) -C sources/go-stable-diffusion libstablediffusion.a
+	CPATH="$(CPATH):/usr/include/opencv4" $(MAKE) -C sources/go-stable-diffusion libstablediffusion.a
 
 ## tiny-dream
 sources/go-tiny-dream:
@@ -263,6 +263,7 @@ dropreplace:
 	$(GOCMD) mod edit -dropreplace github.com/mudler/go-piper
 	$(GOCMD) mod edit -dropreplace github.com/mudler/go-stable-diffusion
 	$(GOCMD) mod edit -dropreplace github.com/nomic-ai/gpt4all/gpt4all-bindings/golang
+	$(GOCMD) mod edit -dropreplace github.com/go-skynet/go-llama.cpp
 
 prepare-sources: get-sources replace
 	$(GOCMD) mod download
@@ -288,16 +289,21 @@ clean: ## Remove build related file
 	rm -rf ./sources
 	rm -rf $(BINARY_NAME)
 	rm -rf release/
-	rm -rf backend-assets
+	rm -rf backend-assets/*
 	$(MAKE) -C backend/cpp/grpc clean
 	$(MAKE) -C backend/cpp/llama clean
 	$(MAKE) dropreplace
+	$(MAKE) protogen-clean
+	rmdir pkg/grpc/proto || true
 
 clean-tests:
 	rm -rf test-models
 	rm -rf test-dir
 	rm -rf core/http/backend-assets
 
+halt-backends:		## Used to clean up stray backends sometimes left running when debugging manually
+	ps | grep 'backend-assets/grpc/' | awk '{print $$1}' | xargs -I {} kill -9 {}
+
 ## Build:
 build: prepare backend-assets grpcs ## Build the project
 	$(info ${GREEN}I local-ai build info:${RESET})
@@ -306,6 +312,12 @@ build: prepare backend-assets grpcs ## Build the project
 	$(info ${GREEN}I LD_FLAGS: ${YELLOW}$(LD_FLAGS)${RESET})
 	CGO_LDFLAGS="$(CGO_LDFLAGS)" $(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o $(BINARY_NAME) ./
 
+build-minimal:
+	BUILD_GRPC_FOR_BACKEND_LLAMA=true GRPC_BACKENDS=backend-assets/grpc/llama-cpp GO_TAGS=none $(MAKE) build
+
+build-api:
+	BUILD_GRPC_FOR_BACKEND_LLAMA=true BUILD_API_ONLY=true GO_TAGS=none $(MAKE) build
+
 dist: build
 	mkdir -p release
 	cp $(BINARY_NAME) release/$(BINARY_NAME)-$(BUILD_ID)-$(OS)-$(ARCH)
@@ -348,7 +360,7 @@ prepare-e2e:
 	mkdir -p $(TEST_DIR)
 	cp -rfv $(abspath ./tests/e2e-fixtures)/gpu.yaml $(TEST_DIR)/gpu.yaml
 	test -e $(TEST_DIR)/ggllm-test-model.bin || wget -q https://huggingface.co/TheBloke/CodeLlama-7B-Instruct-GGUF/resolve/main/codellama-7b-instruct.Q2_K.gguf -O $(TEST_DIR)/ggllm-test-model.bin
-	docker build --build-arg BUILD_GRPC=true --build-arg GRPC_BACKENDS="$(GRPC_BACKENDS)" --build-arg IMAGE_TYPE=core --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg CUDA_MAJOR_VERSION=11 --build-arg CUDA_MINOR_VERSION=7 --build-arg FFMPEG=true -t localai-tests .
+	docker build --build-arg GRPC_BACKENDS="$(GRPC_BACKENDS)" --build-arg IMAGE_TYPE=core --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg CUDA_MAJOR_VERSION=11 --build-arg CUDA_MINOR_VERSION=7 --build-arg FFMPEG=true -t localai-tests .
 
 run-e2e-image:
 	ls -liah $(abspath ./tests/e2e-fixtures)
@@ -356,13 +368,13 @@ run-e2e-image:
 
 run-e2e-aio:
 	@echo 'Running e2e AIO tests'
-	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --flake-attempts 5 -v -r ./tests/e2e-aio
+	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --flake-attempts $(TEST_FLAKES) -v -r ./tests/e2e-aio
 
 test-e2e:
 	@echo 'Running e2e tests'
 	BUILD_TYPE=$(BUILD_TYPE) \
 	LOCALAI_API=http://$(E2E_BRIDGE_IP):5390/v1 \
-	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --flake-attempts 5 -v -r ./tests/e2e
+	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --flake-attempts $(TEST_FLAKES) -v -r ./tests/e2e
 
 teardown-e2e:
 	rm -rf $(TEST_DIR) || true
@@ -370,15 +382,15 @@ teardown-e2e:
 
 test-gpt4all: prepare-test
 	TEST_DIR=$(abspath ./)/test-dir/ FIXTURES=$(abspath ./)/tests/fixtures CONFIG_FILE=$(abspath ./)/test-models/config.yaml MODELS_PATH=$(abspath ./)/test-models \
-	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="gpt4all" --flake-attempts 5 -v -r $(TEST_PATHS)
+	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="gpt4all" --flake-attempts $(TEST_FLAKES) -v -r $(TEST_PATHS)
 
 test-llama: prepare-test
 	TEST_DIR=$(abspath ./)/test-dir/ FIXTURES=$(abspath ./)/tests/fixtures CONFIG_FILE=$(abspath ./)/test-models/config.yaml MODELS_PATH=$(abspath ./)/test-models \
-	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="llama" --flake-attempts 5 -v -r $(TEST_PATHS)
+	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="llama" --flake-attempts $(TEST_FLAKES) -v -r $(TEST_PATHS)
 
 test-llama-gguf: prepare-test
 	TEST_DIR=$(abspath ./)/test-dir/ FIXTURES=$(abspath ./)/tests/fixtures CONFIG_FILE=$(abspath ./)/test-models/config.yaml MODELS_PATH=$(abspath ./)/test-models \
-	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="llama-gguf" --flake-attempts 5 -v -r $(TEST_PATHS)
+	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --label-filter="llama-gguf" --flake-attempts $(TEST_FLAKES) -v -r $(TEST_PATHS)
 
 test-tts: prepare-test
 	TEST_DIR=$(abspath ./)/test-dir/ FIXTURES=$(abspath ./)/tests/fixtures CONFIG_FILE=$(abspath ./)/test-models/config.yaml MODELS_PATH=$(abspath ./)/test-models \
@@ -409,30 +421,144 @@ help: ## Show this help.
 		else if (/^## .*$$/) {printf "  ${CYAN}%s${RESET}\n", substr($$1,4)} \
 		}' $(MAKEFILE_LIST)
 
+.PHONY: protogen
 protogen: protogen-go protogen-python
 
+.PHONY: protogen-clean
+protogen-clean: protogen-go-clean protogen-python-clean
+
+.PHONY: protogen-go
 protogen-go:
+	mkdir -p pkg/grpc/proto
 	protoc -Ibackend/ --go_out=pkg/grpc/proto/ --go_opt=paths=source_relative --go-grpc_out=pkg/grpc/proto/ --go-grpc_opt=paths=source_relative \
     backend/backend.proto
 
-protogen-python:
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/sentencetransformers/ --grpc_python_out=backend/python/sentencetransformers/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/transformers/ --grpc_python_out=backend/python/transformers/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/transformers-musicgen/ --grpc_python_out=backend/python/transformers-musicgen/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/autogptq/ --grpc_python_out=backend/python/autogptq/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/exllama/ --grpc_python_out=backend/python/exllama/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/bark/ --grpc_python_out=backend/python/bark/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/diffusers/ --grpc_python_out=backend/python/diffusers/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/coqui/ --grpc_python_out=backend/python/coqui/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/vall-e-x/ --grpc_python_out=backend/python/vall-e-x/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/vllm/ --grpc_python_out=backend/python/vllm/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/petals/ --grpc_python_out=backend/python/petals/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/mamba/ --grpc_python_out=backend/python/mamba/ backend/backend.proto
-	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/exllama2/ --grpc_python_out=backend/python/exllama2/ backend/backend.proto
+.PHONY: protogen-go-clean
+protogen-go-clean:
+	$(RM) pkg/grpc/proto/backend.pb.go pkg/grpc/proto/backend_grpc.pb.go
+	$(RM) bin/*
+
+.PHONY: protogen-python
+protogen-python: autogptq-protogen bark-protogen coqui-protogen diffusers-protogen exllama-protogen exllama2-protogen mamba-protogen petals-protogen sentencetransformers-protogen transformers-protogen parler-tts-protogen transformers-musicgen-protogen vall-e-x-protogen vllm-protogen
+
+.PHONY: protogen-python-clean
+protogen-python-clean: autogptq-protogen-clean bark-protogen-clean coqui-protogen-clean diffusers-protogen-clean exllama-protogen-clean exllama2-protogen-clean mamba-protogen-clean petals-protogen-clean sentencetransformers-protogen-clean transformers-protogen-clean transformers-musicgen-protogen-clean parler-tts-protogen-clean vall-e-x-protogen-clean vllm-protogen-clean
+
+.PHONY: autogptq-protogen
+autogptq-protogen:
+	$(MAKE) -C backend/python/autogptq protogen
+
+.PHONY: autogptq-protogen-clean
+autogptq-protogen-clean:
+	$(MAKE) -C backend/python/autogptq protogen-clean
+
+.PHONY: bark-protogen
+bark-protogen:
+	$(MAKE) -C backend/python/bark protogen
+
+.PHONY: bark-protogen-clean
+bark-protogen-clean:
+	$(MAKE) -C backend/python/bark protogen-clean
+
+.PHONY: coqui-protogen
+coqui-protogen:
+	$(MAKE) -C backend/python/coqui protogen
+
+.PHONY: coqui-protogen-clean
+coqui-protogen-clean:
+	$(MAKE) -C backend/python/coqui protogen-clean
+
+.PHONY: diffusers-protogen
+diffusers-protogen:
+	$(MAKE) -C backend/python/diffusers protogen
+
+.PHONY: diffusers-protogen-clean
+diffusers-protogen-clean:
+	$(MAKE) -C backend/python/diffusers protogen-clean
+
+.PHONY: exllama-protogen
+exllama-protogen:
+	$(MAKE) -C backend/python/exllama protogen
+
+.PHONY: exllama-protogen-clean
+exllama-protogen-clean:
+	$(MAKE) -C backend/python/exllama protogen-clean
+
+.PHONY: exllama2-protogen
+exllama2-protogen:
+	$(MAKE) -C backend/python/exllama2 protogen
+
+.PHONY: exllama2-protogen-clean
+exllama2-protogen-clean:
+	$(MAKE) -C backend/python/exllama2 protogen-clean
+
+.PHONY: mamba-protogen
+mamba-protogen:
+	$(MAKE) -C backend/python/mamba protogen
+
+.PHONY: mamba-protogen-clean
+mamba-protogen-clean:
+	$(MAKE) -C backend/python/mamba protogen-clean
+
+.PHONY: petals-protogen
+petals-protogen:
+	$(MAKE) -C backend/python/petals protogen
+
+.PHONY: petals-protogen-clean
+petals-protogen-clean:
+	$(MAKE) -C backend/python/petals protogen-clean
+
+.PHONY: sentencetransformers-protogen
+sentencetransformers-protogen:
+	$(MAKE) -C backend/python/sentencetransformers protogen
+
+.PHONY: sentencetransformers-protogen-clean
+sentencetransformers-protogen-clean:
+	$(MAKE) -C backend/python/sentencetransformers protogen-clean
+
+.PHONY: transformers-protogen
+transformers-protogen:
+	$(MAKE) -C backend/python/transformers protogen
+
+.PHONY: transformers-protogen-clean
+transformers-protogen-clean:
+	$(MAKE) -C backend/python/transformers protogen-clean
+
+.PHONY: parler-tts-protogen
+parler-tts-protogen:
+	$(MAKE) -C backend/python/parler-tts protogen
+
+.PHONY: parler-tts-protogen-clean
+parler-tts-protogen-clean:
+	$(MAKE) -C backend/python/parler-tts protogen-clean
+
+.PHONY: transformers-musicgen-protogen
+transformers-musicgen-protogen:
+	$(MAKE) -C backend/python/transformers-musicgen protogen
+
+.PHONY: transformers-musicgen-protogen-clean
+transformers-musicgen-protogen-clean:
+	$(MAKE) -C backend/python/transformers-musicgen protogen-clean
+
+.PHONY: vall-e-x-protogen
+vall-e-x-protogen:
+	$(MAKE) -C backend/python/vall-e-x protogen
+
+.PHONY: vall-e-x-protogen-clean
+vall-e-x-protogen-clean:
+	$(MAKE) -C backend/python/vall-e-x protogen-clean
+
+.PHONY: vllm-protogen
+vllm-protogen:
+	$(MAKE) -C backend/python/vllm protogen
+
+.PHONY: vllm-protogen-clean
+vllm-protogen-clean:
+	$(MAKE) -C backend/python/vllm protogen-clean
 
 ## GRPC
 # Note: it is duplicated in the Dockerfile
-prepare-extra-conda-environments:
+prepare-extra-conda-environments: protogen-python
 	$(MAKE) -C backend/python/autogptq
 	$(MAKE) -C backend/python/bark
 	$(MAKE) -C backend/python/coqui
@@ -442,12 +568,13 @@ prepare-extra-conda-environments:
 	$(MAKE) -C backend/python/sentencetransformers
 	$(MAKE) -C backend/python/transformers
 	$(MAKE) -C backend/python/transformers-musicgen
+	$(MAKE) -C backend/python/parler-tts
 	$(MAKE) -C backend/python/vall-e-x
 	$(MAKE) -C backend/python/exllama
 	$(MAKE) -C backend/python/petals
 	$(MAKE) -C backend/python/exllama2
 
-prepare-test-extra:
+prepare-test-extra: protogen-python
 	$(MAKE) -C backend/python/transformers
 	$(MAKE) -C backend/python/diffusers
 
@@ -471,7 +598,7 @@ backend-assets/gpt4all: sources/gpt4all sources/gpt4all/gpt4all-bindings/golang/
 	@cp sources/gpt4all/gpt4all-bindings/golang/buildllm/*.dylib backend-assets/gpt4all/ || true
 	@cp sources/gpt4all/gpt4all-bindings/golang/buildllm/*.dll backend-assets/gpt4all/ || true
 
-backend-assets/grpc: replace
+backend-assets/grpc: protogen-go replace
 	mkdir -p backend-assets/grpc
 
 backend-assets/grpc/bert-embeddings: sources/go-bert sources/go-bert/libgobert.a backend-assets/grpc
@@ -521,7 +648,10 @@ backend-assets/grpc/llama-ggml: sources/go-llama-ggml sources/go-llama-ggml/libb
 	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(CURDIR)/sources/go-llama-ggml
 	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/go-llama-ggml LIBRARY_PATH=$(CURDIR)/sources/go-llama-ggml \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/llama-ggml ./backend/go/llm/llama-ggml/
-
+#	EXPERIMENTAL:
+ifeq ($(BUILD_TYPE),metal)
+	cp $(CURDIR)/sources/go-llama-ggml/llama.cpp/ggml-metal.metal backend-assets/grpc/
+endif
 backend-assets/grpc/piper: sources/go-piper sources/go-piper/libpiper_binding.a backend-assets/grpc backend-assets/espeak-ng-data
 	CGO_CXXFLAGS="$(PIPER_CGO_CXXFLAGS)" CGO_LDFLAGS="$(PIPER_CGO_LDFLAGS)" LIBRARY_PATH=$(CURDIR)/sources/go-piper \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/piper ./backend/go/tts/
@@ -531,7 +661,7 @@ backend-assets/grpc/rwkv: sources/go-rwkv sources/go-rwkv/librwkv.a backend-asse
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/rwkv ./backend/go/llm/rwkv
 
 backend-assets/grpc/stablediffusion: sources/go-stable-diffusion sources/go-stable-diffusion/libstablediffusion.a backend-assets/grpc
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/go-stable-diffusion/ LIBRARY_PATH=$(CURDIR)/sources/go-stable-diffusion/ \
+	CGO_LDFLAGS="$(CGO_LDFLAGS)" CPATH="$(CPATH):$(CURDIR)/sources/go-stable-diffusion/:/usr/include/opencv4" LIBRARY_PATH=$(CURDIR)/sources/go-stable-diffusion/ \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/stablediffusion ./backend/go/image/stablediffusion
 
 backend-assets/grpc/tinydream: sources/go-tiny-dream sources/go-tiny-dream/libtinydream.a backend-assets/grpc
@@ -556,7 +686,8 @@ docker:
 	docker build \
 		--build-arg BASE_IMAGE=$(BASE_IMAGE) \
 		--build-arg IMAGE_TYPE=$(IMAGE_TYPE) \
-		--build-arg GO_TAGS=$(GO_TAGS) \
+		--build-arg GO_TAGS="$(GO_TAGS)" \
+		--build-arg MAKEFLAGS="$(DOCKER_MAKEFLAGS)" \
 		--build-arg BUILD_TYPE=$(BUILD_TYPE) \
 		-t $(DOCKER_IMAGE) .
 	
@@ -564,6 +695,7 @@ docker-aio:
 	@echo "Building AIO image with base $(BASE_IMAGE) as $(DOCKER_AIO_IMAGE)"
 	docker build \
 		--build-arg BASE_IMAGE=$(BASE_IMAGE) \
+		--build-arg MAKEFLAGS="$(DOCKER_MAKEFLAGS)" \
 		-t $(DOCKER_AIO_IMAGE) -f Dockerfile.aio .
 
 docker-aio-all:
@@ -575,6 +707,7 @@ docker-image-intel:
 		--build-arg BASE_IMAGE=intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04 \
 		--build-arg IMAGE_TYPE=$(IMAGE_TYPE) \
 		--build-arg GO_TAGS="none" \
+		--build-arg MAKEFLAGS="$(DOCKER_MAKEFLAGS)" \
 		--build-arg BUILD_TYPE=sycl_f32 -t $(DOCKER_IMAGE) .
 
 docker-image-intel-xpu:
@@ -582,4 +715,9 @@ docker-image-intel-xpu:
 		--build-arg BASE_IMAGE=intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04 \
 		--build-arg IMAGE_TYPE=$(IMAGE_TYPE) \
 		--build-arg GO_TAGS="none" \
+		--build-arg MAKEFLAGS="$(DOCKER_MAKEFLAGS)" \
 		--build-arg BUILD_TYPE=sycl_f32 -t $(DOCKER_IMAGE) .
+
+.PHONY: swagger
+swagger:
+	swag init -g core/http/api.go --output swagger

README.md

@@ -38,6 +38,10 @@
 </a>
 </p>
 
+> :bulb: Get help - [❓FAQ](https://localai.io/faq/) [💭Discussions](https://github.com/go-skynet/LocalAI/discussions) [:speech_balloon: Discord](https://discord.gg/uJAeKSAGDy) [:book: Documentation website](https://localai.io/)
+>
+> [💻 Quickstart](https://localai.io/basics/getting_started/) [📣 News](https://localai.io/basics/news/) [ 🛫 Examples ](https://github.com/go-skynet/LocalAI/tree/master/examples/) [ 🖼️ Models ](https://localai.io/models/) [ 🚀 Roadmap ](https://github.com/mudler/LocalAI/issues?q=is%3Aissue+is%3Aopen+label%3Aroadmap)
+
 [![tests](https://github.com/go-skynet/LocalAI/actions/workflows/test.yml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/test.yml)[![Build and Release](https://github.com/go-skynet/LocalAI/actions/workflows/release.yaml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/release.yaml)[![build container images](https://github.com/go-skynet/LocalAI/actions/workflows/image.yml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/image.yml)[![Bump dependencies](https://github.com/go-skynet/LocalAI/actions/workflows/bump_deps.yaml/badge.svg)](https://github.com/go-skynet/LocalAI/actions/workflows/bump_deps.yaml)[![Artifact Hub](https://img.shields.io/endpoint?url=https://artifacthub.io/badge/repository/localai)](https://artifacthub.io/packages/search?repo=localai)
 
 **LocalAI** is the free, Open Source OpenAI alternative. LocalAI act as a drop-in replacement REST API that’s compatible with OpenAI (Elevenlabs, Anthropic... ) API specifications for local AI inferencing. It allows you to run LLMs, generate images, audio (and not only) locally or on-prem with consumer grade hardware, supporting multiple model families. Does not require GPU.
@@ -46,18 +50,12 @@
 
 [Roadmap](https://github.com/mudler/LocalAI/issues?q=is%3Aissue+is%3Aopen+label%3Aroadmap)
 
+- Parler-TTS: https://github.com/mudler/LocalAI/pull/2027
+- Landing page: https://github.com/mudler/LocalAI/pull/1922
+- Openvino support: https://github.com/mudler/LocalAI/pull/1892
 - Vector store: https://github.com/mudler/LocalAI/pull/1795
 - All-in-one container image: https://github.com/mudler/LocalAI/issues/1855
-- Parallel function calling: https://github.com/mudler/LocalAI/pull/1726
-- Upload file API: https://github.com/mudler/LocalAI/pull/1703
-- Tools API support: https://github.com/mudler/LocalAI/pull/1715
-- LLaVa 1.6: https://github.com/mudler/LocalAI/pull/1714
-- ROCm container images: https://github.com/mudler/LocalAI/pull/1595
-- Intel GPU support (sycl, transformers, diffusers): https://github.com/mudler/LocalAI/issues/1653
-- Mamba support: https://github.com/mudler/LocalAI/pull/1589
-- Start and share models with config file: https://github.com/mudler/LocalAI/pull/1522
-- 🐸 Coqui: https://github.com/mudler/LocalAI/pull/1489
-- Img2vid https://github.com/mudler/LocalAI/pull/1442
+- Parallel function calling: https://github.com/mudler/LocalAI/pull/1726 / Tools API support: https://github.com/mudler/LocalAI/pull/1715
 
 Hot topics (looking for contributors):
 - Backends v2: https://github.com/mudler/LocalAI/issues/1126
@@ -77,7 +75,7 @@ For those in a hurry, here's a straightforward one-liner to launch a LocalAI AIO
 ```bash
 docker run -ti --name local-ai -p 8080:8080 localai/localai:latest-aio-cpu
 # or, if you have an Nvidia GPU:
-# docker run -ti --name local-ai -p 8080:8080 --gpus all localai/localai:latest-aio-gpu-cuda12
+# docker run -ti --name local-ai -p 8080:8080 --gpus all localai/localai:latest-aio-gpu-nvidia-cuda-12
 ```
 
 ## 🚀 [Features](https://localai.io/features/)

aio/cpu/embeddings.yaml

@@ -1,11 +1,5 @@
-backend: bert-embeddings
-embeddings: true
-f16: true
-
-gpu_layers: 90
-mmap: true
 name: text-embedding-ada-002
-
+backend: bert-embeddings
 parameters:
   model: huggingface://mudler/all-MiniLM-L6-v2/ggml-model-q4_0.bin
 

aio/cpu/image-gen.yaml

@@ -50,4 +50,13 @@ download_files:
   uri: "https://github.com/EdVince/Stable-Diffusion-NCNN/releases/download/naifu/UNetModel-MHA-fp16.bin"
 - filename: "stablediffusion_assets/vocab.txt"
   sha256: "e30e57b6f1e47616982ef898d8922be24e535b4fa3d0110477b3a6f02ebbae7d"
-  uri: "https://raw.githubusercontent.com/EdVince/Stable-Diffusion-NCNN/main/x86/linux/assets/vocab.txt"
+  uri: "https://raw.githubusercontent.com/EdVince/Stable-Diffusion-NCNN/main/x86/linux/assets/vocab.txt"
+
+usage: |
+        curl http://localhost:8080/v1/images/generations \
+          -H "Content-Type: application/json" \
+          -d '{
+            "prompt": "<positive prompt>|<negative prompt>",
+            "step": 25,
+            "size": "512x512"
+          }'

aio/cpu/text-to-text.yaml

@@ -1,25 +1,53 @@
 name: gpt-4
 mmap: true
 parameters:
-  model: huggingface://l3utterfly/phi-2-layla-v1-chatml-gguf/phi-2-layla-v1-chatml-Q8_0.gguf
+  model: huggingface://NousResearch/Hermes-2-Pro-Mistral-7B-GGUF/Hermes-2-Pro-Mistral-7B.Q2_K.gguf
 
 template:
   chat_message: |
-    <|im_start|>{{if eq .RoleName "assistant"}}assistant{{else if eq .RoleName "system"}}system{{else if eq .RoleName "user"}}user{{end}}
-    {{if .Content}}{{.Content}}{{end}}
+    <|im_start|>{{if eq .RoleName "assistant"}}assistant{{else if eq .RoleName "system"}}system{{else if eq .RoleName "tool"}}tool{{else if eq .RoleName "user"}}user{{end}}
+    {{- if .FunctionCall }}<tool_call>{{end}}
+    {{- if eq .RoleName "tool" }}<tool_result>{{end }}
+    {{- if .Content}}
+    {{.Content}}
+    {{- end }}
+    {{- if .FunctionCall}}{{toJson .FunctionCall}}{{end }}
+    {{- if .FunctionCall }}</tool_call>{{end }}
+    {{- if eq .RoleName "tool" }}</tool_result>{{end }}
     <|im_end|>
+  # https://huggingface.co/NousResearch/Hermes-2-Pro-Mistral-7B-GGUF#prompt-format-for-function-calling
+  function: |
+    <|im_start|>system
+    You are a function calling AI model. You are provided with function signatures within <tools></tools> XML tags. You may call one or more functions to assist with the user query. Don't make assumptions about what values to plug into functions. Here are the available tools:
+    <tools>
+    {{range .Functions}}
+    {'type': 'function', 'function': {'name': '{{.Name}}', 'description': '{{.Description}}', 'parameters': {{toJson .Parameters}} }}
+    {{end}}
+    </tools>
+    Use the following pydantic model json schema for each tool call you will make:
+    {'title': 'FunctionCall', 'type': 'object', 'properties': {'arguments': {'title': 'Arguments', 'type': 'object'}, 'name': {'title': 'Name', 'type': 'string'}}, 'required': ['arguments', 'name']}
+    For each function call return a json object with function name and arguments within <tool_call></tool_call> XML tags as follows:
+    <tool_call>
+    {'arguments': <args-dict>, 'name': <function-name>}
+    </tool_call>
+    <|im_end|>
+    {{.Input -}}
+    <|im_start|>assistant
+    <tool_call>
   chat: |
-    {{.Input}}
+    {{.Input -}}
     <|im_start|>assistant
   completion: |
     {{.Input}}
-context_size: 2048
+context_size: 4096
 f16: true
 stopwords:
 - <|im_end|>
 - <dummy32000>
+- "\n</tool_call>"
+- "\n\n\n"
 usage: |
       curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
-          "model": "phi-2-chat",
+          "model": "gpt-4",
           "messages": [{"role": "user", "content": "How are you doing?", "temperature": 0.1}]
       }'

aio/cpu/vision.yaml

@@ -1,8 +1,6 @@
 backend: llama-cpp
 context_size: 4096
 f16: true
-
-gpu_layers: 90
 mmap: true
 name: gpt-4-vision-preview
 
@@ -14,13 +12,6 @@ roles:
 mmproj: bakllava-mmproj.gguf
 parameters:
   model: bakllava.gguf
-  temperature: 0.2
-  top_k: 40
-  top_p: 0.95
-  seed: -1
-mirostat: 2
-mirostat_eta: 1.0
-mirostat_tau: 1.0
 
 template:
   chat: |

aio/entrypoint.sh

@@ -5,53 +5,77 @@ echo "===> LocalAI All-in-One (AIO) container starting..."
 GPU_ACCELERATION=false
 GPU_VENDOR=""
 
+function check_intel() {
+    if lspci | grep -E 'VGA|3D' | grep -iq intel; then
+        echo "Intel GPU detected"
+        if [ -d /opt/intel ]; then
+            GPU_ACCELERATION=true
+            GPU_VENDOR=intel
+        else
+            echo "Intel GPU detected, but Intel GPU drivers are not installed. GPU acceleration will not be available."
+        fi
+    fi
+}
+
+function check_nvidia_wsl() {
+    if lspci | grep -E 'VGA|3D' | grep -iq "Microsoft Corporation Device 008e"; then
+        # We make the assumption this WSL2 cars is NVIDIA, then check for nvidia-smi
+        # Make sure the container was run with `--gpus all` as the only required parameter
+        echo "NVIDIA GPU detected via WSL2"
+        # nvidia-smi should be installed in the container
+        if nvidia-smi; then
+            GPU_ACCELERATION=true
+            GPU_VENDOR=nvidia
+        else
+            echo "NVIDIA GPU detected via WSL2, but nvidia-smi is not installed. GPU acceleration will not be available."
+        fi
+    fi
+}
+
+function check_amd() {
+    if lspci | grep -E 'VGA|3D' | grep -iq amd; then
+        echo "AMD GPU detected"
+        # Check if ROCm is installed
+        if [ -d /opt/rocm ]; then
+            GPU_ACCELERATION=true
+            GPU_VENDOR=amd
+        else
+            echo "AMD GPU detected, but ROCm is not installed. GPU acceleration will not be available."
+        fi
+    fi
+}
+
+function check_nvidia() {
+    if lspci | grep -E 'VGA|3D' | grep -iq nvidia; then
+        echo "NVIDIA GPU detected"
+        # nvidia-smi should be installed in the container
+        if nvidia-smi; then
+            GPU_ACCELERATION=true
+            GPU_VENDOR=nvidia
+        else
+            echo "NVIDIA GPU detected, but nvidia-smi is not installed. GPU acceleration will not be available."
+        fi
+    fi
+}
+
+function check_metal() {
+    if system_profiler SPDisplaysDataType | grep -iq 'Metal'; then
+        echo "Apple Metal supported GPU detected"
+        GPU_ACCELERATION=true
+        GPU_VENDOR=apple
+    fi
+}
+
 function detect_gpu() {
     case "$(uname -s)" in
         Linux)
-            if lspci | grep -E 'VGA|3D' | grep -iq nvidia; then
-                echo "NVIDIA GPU detected"
-                # nvidia-smi should be installed in the container
-                if nvidia-smi; then
-                    GPU_ACCELERATION=true
-                    GPU_VENDOR=nvidia
-                else
-                    echo "NVIDIA GPU detected, but nvidia-smi is not installed. GPU acceleration will not be available."
-                fi
-            elif lspci | grep -E 'VGA|3D' | grep -iq amd; then
-                echo "AMD GPU detected"
-                # Check if ROCm is installed
-                if [ -d /opt/rocm ]; then
-                    GPU_ACCELERATION=true
-                    GPU_VENDOR=amd
-                else
-                    echo "AMD GPU detected, but ROCm is not installed. GPU acceleration will not be available."
-                fi
-            elif lspci | grep -E 'VGA|3D' | grep -iq intel; then
-                echo "Intel GPU detected"
-                if [ -d /opt/intel ]; then
-                    GPU_ACCELERATION=true
-                else
-                    echo "Intel GPU detected, but Intel GPU drivers are not installed. GPU acceleration will not be available."
-                fi
-            elif lspci | grep -E 'VGA|3D' | grep -iq "Microsoft Corporation Device 008e"; then
-                # We make the assumption this WSL2 cars is NVIDIA, then check for nvidia-smi
-                # Make sure the container was run with `--gpus all` as the only required parameter
-                echo "NVIDIA GPU detected via WSL2"
-                # nvidia-smi should be installed in the container
-                if nvidia-smi; then
-                    GPU_ACCELERATION=true
-                    GPU_VENDOR=nvidia
-                else
-                    echo "NVIDIA GPU detected via WSL2, but nvidia-smi is not installed. GPU acceleration will not be available."
-                fi
-            fi
+            check_nvidia
+            check_amd
+            check_intel
+            check_nvidia_wsl
             ;;
         Darwin)
-            if system_profiler SPDisplaysDataType | grep -iq 'Metal'; then
-                echo "Apple Metal supported GPU detected"
-                GPU_ACCELERATION=true
-                GPU_VENDOR=apple
-            fi
+            check_metal
             ;;
     esac
 }
@@ -75,7 +99,8 @@ function detect_gpu_size() {
             echo "Unable to determine NVIDIA GPU memory size. Falling back to CPU."
             GPU_SIZE=gpu-8g
         fi
-
+    elif [ "$GPU_ACCELERATION" = true ] && [ "$GPU_VENDOR" = "intel" ]; then
+        GPU_SIZE=intel
     # Default to a generic GPU size until we implement GPU size detection for non NVIDIA GPUs
     elif [ "$GPU_ACCELERATION" = true ]; then
         echo "Non-NVIDIA GPU detected. Specific GPU memory size detection is not implemented."
@@ -94,8 +119,8 @@ function check_vars() {
         exit 1
     fi
 
-    if [ -z "$SIZE" ]; then
-        echo "SIZE environment variable is not set. Please set it to one of the following: cpu, gpu-8g, gpu-16g, apple"
+    if [ -z "$PROFILE" ]; then
+        echo "PROFILE environment variable is not set. Please set it to one of the following: cpu, gpu-8g, gpu-16g, apple"
         exit 1
     fi
 }
@@ -103,11 +128,11 @@ function check_vars() {
 detect_gpu
 detect_gpu_size
 
-SIZE="${SIZE:-$GPU_SIZE}" # default to cpu
-export MODELS="${MODELS:-/aio/${SIZE}/embeddings.yaml,/aio/${SIZE}/text-to-speech.yaml,/aio/${SIZE}/image-gen.yaml,/aio/${SIZE}/text-to-text.yaml,/aio/${SIZE}/speech-to-text.yaml,/aio/${SIZE}/vision.yaml}"
+PROFILE="${PROFILE:-$GPU_SIZE}" # default to cpu
+export MODELS="${MODELS:-/aio/${PROFILE}/embeddings.yaml,/aio/${PROFILE}/text-to-speech.yaml,/aio/${PROFILE}/image-gen.yaml,/aio/${PROFILE}/text-to-text.yaml,/aio/${PROFILE}/speech-to-text.yaml,/aio/${PROFILE}/vision.yaml}"
 
 check_vars
 
-echo "Starting LocalAI with the following models: $MODELS"
+echo "===> Starting LocalAI[$PROFILE] with the following models: $MODELS"
 
-/build/entrypoint.sh "$@"
+exec /build/entrypoint.sh "$@"

aio/gpu-8g/embeddings.yaml

@@ -1,6 +1,5 @@
 name: text-embedding-ada-002
 backend: sentencetransformers
-embeddings: true
 parameters:
   model: all-MiniLM-L6-v2
 

aio/gpu-8g/image-gen.yaml

@@ -20,7 +20,6 @@ usage: |
           -H "Content-Type: application/json" \
           -d '{
             "prompt": "<positive prompt>|<negative prompt>",
-            "model": "dreamshaper",
             "step": 25,
             "size": "512x512"
           }'

aio/gpu-8g/text-to-text.yaml

@@ -3,39 +3,39 @@ mmap: true
 parameters:
   model: huggingface://NousResearch/Hermes-2-Pro-Mistral-7B-GGUF/Hermes-2-Pro-Mistral-7B.Q6_K.gguf
 
-roles:
-  assistant_function_call: assistant
-  function: tool
 template:
   chat_message: |
-    <|im_start|>{{if eq .RoleName "assistant"}}assistant{{else if eq .RoleName "system"}}system{{else if eq .RoleName "function"}}{{.Role}}{{else if eq .RoleName "user"}}user{{end}}
-    {{ if eq .RoleName "assistant_function_call" }}<tool_call>{{end}}
-    {{ if eq .RoleName "function" }}<tool_result>{{end}}
-    {{if .Content}}{{.Content}}{{end}}
-    {{if .FunctionCall}}{{toJson .FunctionCall}}{{end}}
-    {{ if eq .RoleName "assistant_function_call" }}</tool_call>{{end}}
-    {{ if eq .RoleName "function" }}</tool_result>{{end}}
+    <|im_start|>{{if eq .RoleName "assistant"}}assistant{{else if eq .RoleName "system"}}system{{else if eq .RoleName "tool"}}tool{{else if eq .RoleName "user"}}user{{end}}
+    {{- if .FunctionCall }}<tool_call>{{end}}
+    {{- if eq .RoleName "tool" }}<tool_result>{{end }}
+    {{- if .Content}}
+    {{.Content}}
+    {{- end }}
+    {{- if .FunctionCall}}{{toJson .FunctionCall}}{{end }}
+    {{- if .FunctionCall }}</tool_call>{{end }}
+    {{- if eq .RoleName "tool" }}</tool_result>{{end }}
     <|im_end|>
   # https://huggingface.co/NousResearch/Hermes-2-Pro-Mistral-7B-GGUF#prompt-format-for-function-calling
   function: |
     <|im_start|>system
-    You are a function calling AI model. You are provided with function signatures within <tools></tools> XML tags. You may call one or more functions to assist with the user query. Don't make assumptions about what values to plug into functions. Here are the available tools: 
+    You are a function calling AI model. You are provided with function signatures within <tools></tools> XML tags. You may call one or more functions to assist with the user query. Don't make assumptions about what values to plug into functions. Here are the available tools:
     <tools>
     {{range .Functions}}
     {'type': 'function', 'function': {'name': '{{.Name}}', 'description': '{{.Description}}', 'parameters': {{toJson .Parameters}} }}
     {{end}}
-    </tools> 
-    Use the following pydantic model json schema for each tool call you will make: 
-    {'title': 'FunctionCall', 'type': 'object', 'properties': {'arguments': {'title': 'Arguments', 'type': 'object'}, 'name': {'title': 'Name', 'type': 'string'}}, 'required': ['arguments', 'name']} 
+    </tools>
+    Use the following pydantic model json schema for each tool call you will make:
+    {'title': 'FunctionCall', 'type': 'object', 'properties': {'arguments': {'title': 'Arguments', 'type': 'object'}, 'name': {'title': 'Name', 'type': 'string'}}, 'required': ['arguments', 'name']}
     For each function call return a json object with function name and arguments within <tool_call></tool_call> XML tags as follows:
     <tool_call>
     {'arguments': <args-dict>, 'name': <function-name>}
-    </tool_call><|im_end|>
-    {{.Input}}
+    </tool_call>
+    <|im_end|>
+    {{.Input -}}
     <|im_start|>assistant
     <tool_call>
   chat: |
-    {{.Input}}
+    {{.Input -}}
     <|im_start|>assistant
   completion: |
     {{.Input}}
@@ -44,6 +44,8 @@ f16: true
 stopwords:
 - <|im_end|>
 - <dummy32000>
+- "\n</tool_call>"
+- "\n\n\n"
 usage: |
       curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
           "model": "gpt-4",

aio/gpu-8g/vision.yaml

@@ -1,8 +1,6 @@
 backend: llama-cpp
 context_size: 4096
 f16: true
-
-gpu_layers: 90
 mmap: true
 name: gpt-4-vision-preview
 

aio/intel/embeddings.yaml

@@ -0,0 +1,12 @@
+name: text-embedding-ada-002
+backend: sentencetransformers
+parameters:
+  model: all-MiniLM-L6-v2
+
+usage: |
+    You can test this model with curl like this:
+
+    curl http://localhost:8080/embeddings -X POST -H "Content-Type: application/json" -d '{
+      "input": "Your text string goes here",
+      "model": "text-embedding-ada-002"
+    }'

aio/intel/image-gen.yaml

@@ -0,0 +1,20 @@
+name: stablediffusion
+parameters:
+  model: runwayml/stable-diffusion-v1-5
+backend: diffusers
+step: 25
+f16: true
+diffusers:
+  pipeline_type: StableDiffusionPipeline
+  cuda: true
+  enable_parameters: "negative_prompt,num_inference_steps"
+  scheduler_type: "k_dpmpp_2m"
+
+usage: |
+        curl http://localhost:8080/v1/images/generations \
+          -H "Content-Type: application/json" \
+          -d '{
+            "prompt": "<positive prompt>|<negative prompt>",
+            "step": 25,
+            "size": "512x512"
+          }'

aio/intel/speech-to-text.yaml

@@ -0,0 +1,18 @@
+name: whisper-1
+backend: whisper
+parameters:
+  model: ggml-whisper-base.bin
+
+usage: |
+    ## example audio file
+    wget --quiet --show-progress -O gb1.ogg https://upload.wikimedia.org/wikipedia/commons/1/1f/George_W_Bush_Columbia_FINAL.ogg
+
+    ## Send the example audio file to the transcriptions endpoint
+    curl http://localhost:8080/v1/audio/transcriptions \
+         -H "Content-Type: multipart/form-data" \
+         -F file="@$PWD/gb1.ogg" -F model="whisper-1"
+
+download_files:
+- filename: "ggml-whisper-base.bin"
+  sha256: "60ed5bc3dd14eea856493d334349b405782ddcaf0028d4b5df4088345fba2efe"
+  uri: "https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-base.bin"

aio/intel/text-to-speech.yaml

@@ -0,0 +1,15 @@
+name: tts-1
+download_files:
+  - filename: voice-en-us-amy-low.tar.gz
+    uri: https://github.com/rhasspy/piper/releases/download/v0.0.2/voice-en-us-amy-low.tar.gz
+
+parameters:
+  model: en-us-amy-low.onnx
+
+usage: |
+    To test if this model works as expected, you can use the following curl command:
+
+    curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{
+      "model":"tts-1",
+      "input": "Hi, this is a test."
+    }'

aio/intel/text-to-text.yaml

@@ -0,0 +1,53 @@
+name: gpt-4
+mmap: false
+f16: false
+parameters:
+  model: huggingface://NousResearch/Hermes-2-Pro-Mistral-7B-GGUF/Hermes-2-Pro-Mistral-7B.Q6_K.gguf
+
+template:
+  chat_message: |
+    <|im_start|>{{if eq .RoleName "assistant"}}assistant{{else if eq .RoleName "system"}}system{{else if eq .RoleName "tool"}}tool{{else if eq .RoleName "user"}}user{{end}}
+    {{- if .FunctionCall }}<tool_call>{{end}}
+    {{- if eq .RoleName "tool" }}<tool_result>{{end }}
+    {{- if .Content}}
+    {{.Content}}
+    {{- end }}
+    {{- if .FunctionCall}}{{toJson .FunctionCall}}{{end }}
+    {{- if .FunctionCall }}</tool_call>{{end }}
+    {{- if eq .RoleName "tool" }}</tool_result>{{end }}
+    <|im_end|>
+  # https://huggingface.co/NousResearch/Hermes-2-Pro-Mistral-7B-GGUF#prompt-format-for-function-calling
+  function: |
+    <|im_start|>system
+    You are a function calling AI model. You are provided with function signatures within <tools></tools> XML tags. You may call one or more functions to assist with the user query. Don't make assumptions about what values to plug into functions. Here are the available tools:
+    <tools>
+    {{range .Functions}}
+    {'type': 'function', 'function': {'name': '{{.Name}}', 'description': '{{.Description}}', 'parameters': {{toJson .Parameters}} }}
+    {{end}}
+    </tools>
+    Use the following pydantic model json schema for each tool call you will make:
+    {'title': 'FunctionCall', 'type': 'object', 'properties': {'arguments': {'title': 'Arguments', 'type': 'object'}, 'name': {'title': 'Name', 'type': 'string'}}, 'required': ['arguments', 'name']}
+    For each function call return a json object with function name and arguments within <tool_call></tool_call> XML tags as follows:
+    <tool_call>
+    {'arguments': <args-dict>, 'name': <function-name>}
+    </tool_call>
+    <|im_end|>
+    {{.Input -}}
+    <|im_start|>assistant
+    <tool_call>
+  chat: |
+    {{.Input -}}
+    <|im_start|>assistant
+  completion: |
+    {{.Input}}
+context_size: 4096
+stopwords:
+- <|im_end|>
+- "\n</tool_call>"
+- <dummy32000>
+- "\n\n\n"
+usage: |
+      curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
+          "model": "gpt-4",
+          "messages": [{"role": "user", "content": "How are you doing?", "temperature": 0.1}]
+      }'

aio/intel/vision.yaml

@@ -0,0 +1,35 @@
+backend: llama-cpp
+context_size: 4096
+mmap: false
+f16: false
+name: gpt-4-vision-preview
+
+roles:
+  user: "USER:"
+  assistant: "ASSISTANT:"
+  system: "SYSTEM:"
+
+mmproj: llava-v1.6-7b-mmproj-f16.gguf
+parameters:
+  model: llava-v1.6-mistral-7b.Q5_K_M.gguf
+  temperature: 0.2
+  top_k: 40
+  top_p: 0.95
+  seed: -1
+
+template:
+  chat: |
+    A chat between a curious human and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the human's questions.
+    {{.Input}}
+    ASSISTANT:
+
+download_files:
+- filename: llava-v1.6-mistral-7b.Q5_K_M.gguf
+  uri: huggingface://cjpais/llava-1.6-mistral-7b-gguf/llava-v1.6-mistral-7b.Q5_K_M.gguf
+- filename: llava-v1.6-7b-mmproj-f16.gguf
+  uri: huggingface://cjpais/llava-1.6-mistral-7b-gguf/mmproj-model-f16.gguf
+
+usage: |
+    curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
+        "model": "gpt-4-vision-preview",
+        "messages": [{"role": "user", "content": [{"type":"text", "text": "What is in the image?"}, {"type": "image_url", "image_url": {"url": "https://upload.wikimedia.org/wikipedia/commons/thumb/d/dd/Gfp-wisconsin-madison-the-nature-boardwalk.jpg/2560px-Gfp-wisconsin-madison-the-nature-boardwalk.jpg" }}], "temperature": 0.9}]}'

backend/backend.proto

@@ -107,11 +107,15 @@ message PredictOptions {
   string NegativePrompt = 40;
   int32 NDraft = 41;
   repeated string Images = 42;
+  bool UseTokenizerTemplate = 43;
+  repeated Message Messages = 44;
 }
 
 // The response message containing the result
 message Reply {
   bytes message = 1;
+  int32 tokens = 2;
+  int32 prompt_tokens = 3;
 }
 
 message ModelOptions {
@@ -256,3 +260,8 @@ message StatusResponse {
   State state = 1;
   MemoryUsageData memory = 2;
 }
+
+message Message {
+  string role = 1;
+  string content = 2;
+}

backend/backend_grpc.pb.go

@@ -1,457 +0,0 @@
-// Code generated by protoc-gen-go-grpc. DO NOT EDIT.
-// versions:
-// - protoc-gen-go-grpc v1.2.0
-// - protoc             v4.23.4
-// source: backend/backend.proto
-
-package proto
-
-import (
-	context "context"
-	grpc "google.golang.org/grpc"
-	codes "google.golang.org/grpc/codes"
-	status "google.golang.org/grpc/status"
-)
-
-// This is a compile-time assertion to ensure that this generated file
-// is compatible with the grpc package it is being compiled against.
-// Requires gRPC-Go v1.32.0 or later.
-const _ = grpc.SupportPackageIsVersion7
-
-// BackendClient is the client API for Backend service.
-//
-// For semantics around ctx use and closing/ending streaming RPCs, please refer to https://pkg.go.dev/google.golang.org/grpc/?tab=doc#ClientConn.NewStream.
-type BackendClient interface {
-	Health(ctx context.Context, in *HealthMessage, opts ...grpc.CallOption) (*Reply, error)
-	Predict(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*Reply, error)
-	LoadModel(ctx context.Context, in *ModelOptions, opts ...grpc.CallOption) (*Result, error)
-	PredictStream(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (Backend_PredictStreamClient, error)
-	Embedding(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*EmbeddingResult, error)
-	GenerateImage(ctx context.Context, in *GenerateImageRequest, opts ...grpc.CallOption) (*Result, error)
-	AudioTranscription(ctx context.Context, in *TranscriptRequest, opts ...grpc.CallOption) (*TranscriptResult, error)
-	TTS(ctx context.Context, in *TTSRequest, opts ...grpc.CallOption) (*Result, error)
-	TokenizeString(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*TokenizationResponse, error)
-	Status(ctx context.Context, in *HealthMessage, opts ...grpc.CallOption) (*StatusResponse, error)
-}
-
-type backendClient struct {
-	cc grpc.ClientConnInterface
-}
-
-func NewBackendClient(cc grpc.ClientConnInterface) BackendClient {
-	return &backendClient{cc}
-}
-
-func (c *backendClient) Health(ctx context.Context, in *HealthMessage, opts ...grpc.CallOption) (*Reply, error) {
-	out := new(Reply)
-	err := c.cc.Invoke(ctx, "/backend.Backend/Health", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}
-
-func (c *backendClient) Predict(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*Reply, error) {
-	out := new(Reply)
-	err := c.cc.Invoke(ctx, "/backend.Backend/Predict", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}
-
-func (c *backendClient) LoadModel(ctx context.Context, in *ModelOptions, opts ...grpc.CallOption) (*Result, error) {
-	out := new(Result)
-	err := c.cc.Invoke(ctx, "/backend.Backend/LoadModel", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}
-
-func (c *backendClient) PredictStream(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (Backend_PredictStreamClient, error) {
-	stream, err := c.cc.NewStream(ctx, &Backend_ServiceDesc.Streams[0], "/backend.Backend/PredictStream", opts...)
-	if err != nil {
-		return nil, err
-	}
-	x := &backendPredictStreamClient{stream}
-	if err := x.ClientStream.SendMsg(in); err != nil {
-		return nil, err
-	}
-	if err := x.ClientStream.CloseSend(); err != nil {
-		return nil, err
-	}
-	return x, nil
-}
-
-type Backend_PredictStreamClient interface {
-	Recv() (*Reply, error)
-	grpc.ClientStream
-}
-
-type backendPredictStreamClient struct {
-	grpc.ClientStream
-}
-
-func (x *backendPredictStreamClient) Recv() (*Reply, error) {
-	m := new(Reply)
-	if err := x.ClientStream.RecvMsg(m); err != nil {
-		return nil, err
-	}
-	return m, nil
-}
-
-func (c *backendClient) Embedding(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*EmbeddingResult, error) {
-	out := new(EmbeddingResult)
-	err := c.cc.Invoke(ctx, "/backend.Backend/Embedding", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}
-
-func (c *backendClient) GenerateImage(ctx context.Context, in *GenerateImageRequest, opts ...grpc.CallOption) (*Result, error) {
-	out := new(Result)
-	err := c.cc.Invoke(ctx, "/backend.Backend/GenerateImage", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}
-
-func (c *backendClient) AudioTranscription(ctx context.Context, in *TranscriptRequest, opts ...grpc.CallOption) (*TranscriptResult, error) {
-	out := new(TranscriptResult)
-	err := c.cc.Invoke(ctx, "/backend.Backend/AudioTranscription", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}
-
-func (c *backendClient) TTS(ctx context.Context, in *TTSRequest, opts ...grpc.CallOption) (*Result, error) {
-	out := new(Result)
-	err := c.cc.Invoke(ctx, "/backend.Backend/TTS", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}
-
-func (c *backendClient) TokenizeString(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*TokenizationResponse, error) {
-	out := new(TokenizationResponse)
-	err := c.cc.Invoke(ctx, "/backend.Backend/TokenizeString", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}
-
-func (c *backendClient) Status(ctx context.Context, in *HealthMessage, opts ...grpc.CallOption) (*StatusResponse, error) {
-	out := new(StatusResponse)
-	err := c.cc.Invoke(ctx, "/backend.Backend/Status", in, out, opts...)
-	if err != nil {
-		return nil, err
-	}
-	return out, nil
-}
-
-// BackendServer is the server API for Backend service.
-// All implementations must embed UnimplementedBackendServer
-// for forward compatibility
-type BackendServer interface {
-	Health(context.Context, *HealthMessage) (*Reply, error)
-	Predict(context.Context, *PredictOptions) (*Reply, error)
-	LoadModel(context.Context, *ModelOptions) (*Result, error)
-	PredictStream(*PredictOptions, Backend_PredictStreamServer) error
-	Embedding(context.Context, *PredictOptions) (*EmbeddingResult, error)
-	GenerateImage(context.Context, *GenerateImageRequest) (*Result, error)
-	AudioTranscription(context.Context, *TranscriptRequest) (*TranscriptResult, error)
-	TTS(context.Context, *TTSRequest) (*Result, error)
-	TokenizeString(context.Context, *PredictOptions) (*TokenizationResponse, error)
-	Status(context.Context, *HealthMessage) (*StatusResponse, error)
-	mustEmbedUnimplementedBackendServer()
-}
-
-// UnimplementedBackendServer must be embedded to have forward compatible implementations.
-type UnimplementedBackendServer struct {
-}
-
-func (UnimplementedBackendServer) Health(context.Context, *HealthMessage) (*Reply, error) {
-	return nil, status.Errorf(codes.Unimplemented, "method Health not implemented")
-}
-func (UnimplementedBackendServer) Predict(context.Context, *PredictOptions) (*Reply, error) {
-	return nil, status.Errorf(codes.Unimplemented, "method Predict not implemented")
-}
-func (UnimplementedBackendServer) LoadModel(context.Context, *ModelOptions) (*Result, error) {
-	return nil, status.Errorf(codes.Unimplemented, "method LoadModel not implemented")
-}
-func (UnimplementedBackendServer) PredictStream(*PredictOptions, Backend_PredictStreamServer) error {
-	return status.Errorf(codes.Unimplemented, "method PredictStream not implemented")
-}
-func (UnimplementedBackendServer) Embedding(context.Context, *PredictOptions) (*EmbeddingResult, error) {
-	return nil, status.Errorf(codes.Unimplemented, "method Embedding not implemented")
-}
-func (UnimplementedBackendServer) GenerateImage(context.Context, *GenerateImageRequest) (*Result, error) {
-	return nil, status.Errorf(codes.Unimplemented, "method GenerateImage not implemented")
-}
-func (UnimplementedBackendServer) AudioTranscription(context.Context, *TranscriptRequest) (*TranscriptResult, error) {
-	return nil, status.Errorf(codes.Unimplemented, "method AudioTranscription not implemented")
-}
-func (UnimplementedBackendServer) TTS(context.Context, *TTSRequest) (*Result, error) {
-	return nil, status.Errorf(codes.Unimplemented, "method TTS not implemented")
-}
-func (UnimplementedBackendServer) TokenizeString(context.Context, *PredictOptions) (*TokenizationResponse, error) {
-	return nil, status.Errorf(codes.Unimplemented, "method TokenizeString not implemented")
-}
-func (UnimplementedBackendServer) Status(context.Context, *HealthMessage) (*StatusResponse, error) {
-	return nil, status.Errorf(codes.Unimplemented, "method Status not implemented")
-}
-func (UnimplementedBackendServer) mustEmbedUnimplementedBackendServer() {}
-
-// UnsafeBackendServer may be embedded to opt out of forward compatibility for this service.
-// Use of this interface is not recommended, as added methods to BackendServer will
-// result in compilation errors.
-type UnsafeBackendServer interface {
-	mustEmbedUnimplementedBackendServer()
-}
-
-func RegisterBackendServer(s grpc.ServiceRegistrar, srv BackendServer) {
-	s.RegisterService(&Backend_ServiceDesc, srv)
-}
-
-func _Backend_Health_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
-	in := new(HealthMessage)
-	if err := dec(in); err != nil {
-		return nil, err
-	}
-	if interceptor == nil {
-		return srv.(BackendServer).Health(ctx, in)
-	}
-	info := &grpc.UnaryServerInfo{
-		Server:     srv,
-		FullMethod: "/backend.Backend/Health",
-	}
-	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
-		return srv.(BackendServer).Health(ctx, req.(*HealthMessage))
-	}
-	return interceptor(ctx, in, info, handler)
-}
-
-func _Backend_Predict_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
-	in := new(PredictOptions)
-	if err := dec(in); err != nil {
-		return nil, err
-	}
-	if interceptor == nil {
-		return srv.(BackendServer).Predict(ctx, in)
-	}
-	info := &grpc.UnaryServerInfo{
-		Server:     srv,
-		FullMethod: "/backend.Backend/Predict",
-	}
-	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
-		return srv.(BackendServer).Predict(ctx, req.(*PredictOptions))
-	}
-	return interceptor(ctx, in, info, handler)
-}
-
-func _Backend_LoadModel_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
-	in := new(ModelOptions)
-	if err := dec(in); err != nil {
-		return nil, err
-	}
-	if interceptor == nil {
-		return srv.(BackendServer).LoadModel(ctx, in)
-	}
-	info := &grpc.UnaryServerInfo{
-		Server:     srv,
-		FullMethod: "/backend.Backend/LoadModel",
-	}
-	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
-		return srv.(BackendServer).LoadModel(ctx, req.(*ModelOptions))
-	}
-	return interceptor(ctx, in, info, handler)
-}
-
-func _Backend_PredictStream_Handler(srv interface{}, stream grpc.ServerStream) error {
-	m := new(PredictOptions)
-	if err := stream.RecvMsg(m); err != nil {
-		return err
-	}
-	return srv.(BackendServer).PredictStream(m, &backendPredictStreamServer{stream})
-}
-
-type Backend_PredictStreamServer interface {
-	Send(*Reply) error
-	grpc.ServerStream
-}
-
-type backendPredictStreamServer struct {
-	grpc.ServerStream
-}
-
-func (x *backendPredictStreamServer) Send(m *Reply) error {
-	return x.ServerStream.SendMsg(m)
-}
-
-func _Backend_Embedding_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
-	in := new(PredictOptions)
-	if err := dec(in); err != nil {
-		return nil, err
-	}
-	if interceptor == nil {
-		return srv.(BackendServer).Embedding(ctx, in)
-	}
-	info := &grpc.UnaryServerInfo{
-		Server:     srv,
-		FullMethod: "/backend.Backend/Embedding",
-	}
-	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
-		return srv.(BackendServer).Embedding(ctx, req.(*PredictOptions))
-	}
-	return interceptor(ctx, in, info, handler)
-}
-
-func _Backend_GenerateImage_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
-	in := new(GenerateImageRequest)
-	if err := dec(in); err != nil {
-		return nil, err
-	}
-	if interceptor == nil {
-		return srv.(BackendServer).GenerateImage(ctx, in)
-	}
-	info := &grpc.UnaryServerInfo{
-		Server:     srv,
-		FullMethod: "/backend.Backend/GenerateImage",
-	}
-	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
-		return srv.(BackendServer).GenerateImage(ctx, req.(*GenerateImageRequest))
-	}
-	return interceptor(ctx, in, info, handler)
-}
-
-func _Backend_AudioTranscription_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
-	in := new(TranscriptRequest)
-	if err := dec(in); err != nil {
-		return nil, err
-	}
-	if interceptor == nil {
-		return srv.(BackendServer).AudioTranscription(ctx, in)
-	}
-	info := &grpc.UnaryServerInfo{
-		Server:     srv,
-		FullMethod: "/backend.Backend/AudioTranscription",
-	}
-	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
-		return srv.(BackendServer).AudioTranscription(ctx, req.(*TranscriptRequest))
-	}
-	return interceptor(ctx, in, info, handler)
-}
-
-func _Backend_TTS_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
-	in := new(TTSRequest)
-	if err := dec(in); err != nil {
-		return nil, err
-	}
-	if interceptor == nil {
-		return srv.(BackendServer).TTS(ctx, in)
-	}
-	info := &grpc.UnaryServerInfo{
-		Server:     srv,
-		FullMethod: "/backend.Backend/TTS",
-	}
-	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
-		return srv.(BackendServer).TTS(ctx, req.(*TTSRequest))
-	}
-	return interceptor(ctx, in, info, handler)
-}
-
-func _Backend_TokenizeString_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
-	in := new(PredictOptions)
-	if err := dec(in); err != nil {
-		return nil, err
-	}
-	if interceptor == nil {
-		return srv.(BackendServer).TokenizeString(ctx, in)
-	}
-	info := &grpc.UnaryServerInfo{
-		Server:     srv,
-		FullMethod: "/backend.Backend/TokenizeString",
-	}
-	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
-		return srv.(BackendServer).TokenizeString(ctx, req.(*PredictOptions))
-	}
-	return interceptor(ctx, in, info, handler)
-}
-
-func _Backend_Status_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
-	in := new(HealthMessage)
-	if err := dec(in); err != nil {
-		return nil, err
-	}
-	if interceptor == nil {
-		return srv.(BackendServer).Status(ctx, in)
-	}
-	info := &grpc.UnaryServerInfo{
-		Server:     srv,
-		FullMethod: "/backend.Backend/Status",
-	}
-	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
-		return srv.(BackendServer).Status(ctx, req.(*HealthMessage))
-	}
-	return interceptor(ctx, in, info, handler)
-}
-
-// Backend_ServiceDesc is the grpc.ServiceDesc for Backend service.
-// It's only intended for direct use with grpc.RegisterService,
-// and not to be introspected or modified (even as a copy)
-var Backend_ServiceDesc = grpc.ServiceDesc{
-	ServiceName: "backend.Backend",
-	HandlerType: (*BackendServer)(nil),
-	Methods: []grpc.MethodDesc{
-		{
-			MethodName: "Health",
-			Handler:    _Backend_Health_Handler,
-		},
-		{
-			MethodName: "Predict",
-			Handler:    _Backend_Predict_Handler,
-		},
-		{
-			MethodName: "LoadModel",
-			Handler:    _Backend_LoadModel_Handler,
-		},
-		{
-			MethodName: "Embedding",
-			Handler:    _Backend_Embedding_Handler,
-		},
-		{
-			MethodName: "GenerateImage",
-			Handler:    _Backend_GenerateImage_Handler,
-		},
-		{
-			MethodName: "AudioTranscription",
-			Handler:    _Backend_AudioTranscription_Handler,
-		},
-		{
-			MethodName: "TTS",
-			Handler:    _Backend_TTS_Handler,
-		},
-		{
-			MethodName: "TokenizeString",
-			Handler:    _Backend_TokenizeString_Handler,
-		},
-		{
-			MethodName: "Status",
-			Handler:    _Backend_Status_Handler,
-		},
-	},
-	Streams: []grpc.StreamDesc{
-		{
-			StreamName:    "PredictStream",
-			Handler:       _Backend_PredictStream_Handler,
-			ServerStreams: true,
-		},
-	},
-	Metadata: "backend/backend.proto",
-}

backend/cpp/grpc/Makefile

@@ -5,7 +5,6 @@ SYSTEM ?= $(HOST_SYSTEM)
 TAG_LIB_GRPC?=v1.59.0
 GIT_REPO_LIB_GRPC?=https://github.com/grpc/grpc.git
 GIT_CLONE_DEPTH?=1
-NUM_BUILD_THREADS?=$(shell nproc --ignore=1)
 
 INSTALLED_PACKAGES=installed_packages
 GRPC_REPO=grpc_repo
@@ -52,7 +51,7 @@ $(GRPC_REPO):
 
 $(GRPC_BUILD): $(GRPC_REPO)
 	mkdir -p $(GRPC_BUILD)
-	cd $(GRPC_BUILD) && cmake $(CMAKE_ARGS) ../$(GRPC_REPO)/grpc && cmake --build . -- -j ${NUM_BUILD_THREADS} && cmake --build . --target install -- -j ${NUM_BUILD_THREADS}
+	cd $(GRPC_BUILD) && cmake $(CMAKE_ARGS) ../$(GRPC_REPO)/grpc && cmake --build . && cmake --build . --target install
 
 build: $(INSTALLED_PACKAGES)
 

backend/cpp/llama/grpc-server.cpp

@@ -2332,6 +2332,10 @@ public:
                 std::string completion_text = result.result_json.value("content", "");
 
                 reply.set_message(completion_text);
+                int32_t tokens_predicted = result.result_json.value("tokens_predicted", 0);
+                reply.set_tokens(tokens_predicted);
+                int32_t tokens_evaluated = result.result_json.value("tokens_evaluated", 0);
+                reply.set_prompt_tokens(tokens_evaluated);
 
                 // Send the reply
                 writer->Write(reply);
@@ -2357,6 +2361,10 @@ public:
         task_result result = llama.queue_results.recv(task_id);
         if (!result.error && result.stop) {
             completion_text = result.result_json.value("content", "");
+            int32_t tokens_predicted = result.result_json.value("tokens_predicted", 0);
+            int32_t tokens_evaluated = result.result_json.value("tokens_evaluated", 0);
+            reply->set_prompt_tokens(tokens_evaluated);
+            reply->set_tokens(tokens_predicted);
             reply->set_message(completion_text);
         }
         else

backend/go/transcribe/transcript.go

@@ -21,7 +21,7 @@ func runCommand(command []string) (string, error) {
 // AudioToWav converts audio to wav for transcribe.
 // TODO: use https://github.com/mccoyst/ogg?
 func audioToWav(src, dst string) error {
-    command := []string{"ffmpeg", "-i", src, "-format", "s16le", "-ar", "16000", "-ac", "1", "-acodec", "pcm_s16le", dst}
+	command := []string{"ffmpeg", "-i", src, "-format", "s16le", "-ar", "16000", "-ac", "1", "-acodec", "pcm_s16le", dst}
 	out, err := runCommand(command)
 	if err != nil {
 		return fmt.Errorf("error: %w out: %s", err, out)
@@ -29,8 +29,8 @@ func audioToWav(src, dst string) error {
 	return nil
 }
 
-func Transcript(model whisper.Model, audiopath, language string, threads uint) (schema.Result, error) {
-	res := schema.Result{}
+func Transcript(model whisper.Model, audiopath, language string, threads uint) (schema.TranscriptionResult, error) {
+	res := schema.TranscriptionResult{}
 
 	dir, err := os.MkdirTemp("", "whisper")
 	if err != nil {

backend/go/transcribe/whisper.go

@@ -21,6 +21,6 @@ func (sd *Whisper) Load(opts *pb.ModelOptions) error {
 	return err
 }
 
-func (sd *Whisper) AudioTranscription(opts *pb.TranscriptRequest) (schema.Result, error) {
+func (sd *Whisper) AudioTranscription(opts *pb.TranscriptRequest) (schema.TranscriptionResult, error) {
 	return Transcript(sd.whisper, opts.Dst, opts.Language, uint(opts.Threads))
 }

backend/python/autogptq/Makefile

@@ -1,4 +1,13 @@
 .PHONY: autogptq
-autogptq:
+autogptq: protogen
 	$(MAKE) -C ../common-env/transformers
 
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/autogptq/autogptq.py

@@ -5,12 +5,14 @@ import signal
 import sys
 import os
 import time
+import base64
 
 import grpc
 import backend_pb2
 import backend_pb2_grpc
+
 from auto_gptq import AutoGPTQForCausalLM
-from transformers import AutoTokenizer
+from transformers import AutoTokenizer, AutoModelForCausalLM
 from transformers import TextGenerationPipeline
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
@@ -28,9 +30,18 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             if request.Device != "":
                 device = request.Device
 
-            tokenizer = AutoTokenizer.from_pretrained(request.Model, use_fast=request.UseFastTokenizer)
+            # support loading local model files
+            model_path = os.path.join(os.environ.get('MODELS_PATH', './'), request.Model)
+            tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=True, trust_remote_code=request.TrustRemoteCode)
 
-            model = AutoGPTQForCausalLM.from_quantized(request.Model,
+            # support model `Qwen/Qwen-VL-Chat-Int4`
+            if "qwen-vl" in request.Model.lower():
+                self.model_name = "Qwen-VL-Chat"
+                model = AutoModelForCausalLM.from_pretrained(model_path, 
+                    trust_remote_code=request.TrustRemoteCode,
+                    device_map="auto").eval()
+            else:
+                model = AutoGPTQForCausalLM.from_quantized(model_path,
                     model_basename=request.ModelBaseName,
                     use_safetensors=True,
                     trust_remote_code=request.TrustRemoteCode,
@@ -55,6 +66,11 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         if request.TopP != 0.0:
             top_p = request.TopP
 
+        
+        prompt_images = self.recompile_vl_prompt(request)
+        compiled_prompt = prompt_images[0]
+        print(f"Prompt: {compiled_prompt}", file=sys.stderr)
+
         # Implement Predict RPC
         pipeline = TextGenerationPipeline(
             model=self.model, 
@@ -64,10 +80,17 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             top_p=top_p,
             repetition_penalty=penalty,
             )
-        t = pipeline(request.Prompt)[0]["generated_text"]
-        # Remove prompt from response if present
-        if request.Prompt in t:
-            t = t.replace(request.Prompt, "")
+        t = pipeline(compiled_prompt)[0]["generated_text"]
+        print(f"generated_text: {t}", file=sys.stderr)
+        
+        if compiled_prompt in t:
+            t = t.replace(compiled_prompt, "")
+        # house keeping. Remove the image files from /tmp folder
+        for img_path in prompt_images[1]:
+            try:
+                os.remove(img_path)
+            except Exception as e:
+                print(f"Error removing image file: {img_path}, {e}", file=sys.stderr)
 
         return backend_pb2.Result(message=bytes(t, encoding='utf-8'))
 
@@ -78,6 +101,24 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         # Not implemented yet
         return self.Predict(request, context)
 
+    def recompile_vl_prompt(self, request):
+        prompt = request.Prompt
+        image_paths = []
+
+        if "qwen-vl" in self.model_name.lower():
+            # request.Images is an array which contains base64 encoded images. Iterate the request.Images array, decode and save each image to /tmp folder with a random filename.
+            # Then, save the image file paths to an array "image_paths".
+            # read "request.Prompt", replace "[img-%d]" with the image file paths in the order they appear in "image_paths". Save the new prompt to "prompt".
+            for i, img in enumerate(request.Images):
+                timestamp = str(int(time.time() * 1000))  # Generate timestamp
+                img_path = f"/tmp/vl-{timestamp}.jpg"  # Use timestamp in filename
+                with open(img_path, "wb") as f:
+                    f.write(base64.b64decode(img))
+                image_paths.append(img_path)
+                prompt = prompt.replace(f"[img-{i}]", "<img>" + img_path + "</img>,")
+        else:
+            prompt = request.Prompt
+        return (prompt, image_paths)
 
 def serve(address):
     server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))

backend/python/autogptq/autogptq.yml

@@ -1,3 +1,7 @@
+####
+# Attention! This file is abandoned. 
+# Please use the ../common-env/transformers/transformers.yml file to manage dependencies.
+###
 name: autogptq
 channels:
   - defaults
@@ -24,12 +28,12 @@ dependencies:
   - xz=5.4.2=h5eee18b_0
   - zlib=1.2.13=h5eee18b_0
   - pip:
-      - accelerate==0.23.0
+      - accelerate==0.27.0
       - aiohttp==3.8.5
       - aiosignal==1.3.1
       - async-timeout==4.0.3
       - attrs==23.1.0
-      - auto-gptq==0.4.2
+      - auto-gptq==0.7.1
       - certifi==2023.7.22
       - charset-normalizer==3.3.0
       - datasets==2.14.5
@@ -59,6 +63,7 @@ dependencies:
       - nvidia-nccl-cu12==2.18.1
       - nvidia-nvjitlink-cu12==12.2.140
       - nvidia-nvtx-cu12==12.1.105
+      - optimum==1.17.1
       - packaging==23.2
       - pandas==2.1.1
       - peft==0.5.0
@@ -75,9 +80,11 @@ dependencies:
       - six==1.16.0
       - sympy==1.12
       - tokenizers==0.14.0
-      - torch==2.1.0
       - tqdm==4.66.1
+      - torch==2.2.1
+      - torchvision==0.17.1
       - transformers==4.34.0
+      - transformers_stream_generator==0.0.5
       - triton==2.1.0
       - typing-extensions==4.8.0
       - tzdata==2023.3

backend/python/autogptq/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/bark/Makefile

@@ -1,15 +1,25 @@
 .PHONY: ttsbark
-ttsbark:
+ttsbark: protogen
 	$(MAKE) -C ../common-env/transformers
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running bark..."
 	bash run.sh
 	@echo "bark run."
 
 .PHONY: test
-test:
+test: protogen
 	@echo "Testing bark..."
 	bash test.sh
 	@echo "bark tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/bark/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/common-env/transformers/install.sh

@@ -2,6 +2,7 @@
 set -ex
 
 SKIP_CONDA=${SKIP_CONDA:-0}
+REQUIREMENTS_FILE=$1
 
 # Check if environment exist
 conda_env_exists(){
@@ -14,7 +15,7 @@ else
     export PATH=$PATH:/opt/conda/bin
     if conda_env_exists "transformers" ; then
         echo "Creating virtual environment..."
-        conda env create --name transformers --file $1
+        conda env create --name transformers --file $REQUIREMENTS_FILE
         echo "Virtual environment created."
     else 
         echo "Virtual environment already exists."
@@ -25,14 +26,19 @@ if [ -d "/opt/intel" ]; then
     # Intel GPU: If the directory exists, we assume we are using the intel image
     # (no conda env)
     # https://github.com/intel/intel-extension-for-pytorch/issues/538
-    pip install intel-extension-for-transformers datasets sentencepiece tiktoken neural_speed
+    pip install intel-extension-for-transformers datasets sentencepiece tiktoken neural_speed optimum[openvino]
+fi
+
+# If we didn't skip conda, activate the environment
+# to install FlashAttention
+if [ $SKIP_CONDA -eq 0 ]; then
+    source activate transformers
+fi
+if [[ $REQUIREMENTS_FILE =~ -nvidia.yml$ ]]; then
+    #TODO: FlashAttention is supported on nvidia and ROCm, but ROCm install can't be done this easily
+    pip install flash-attn --no-build-isolation
 fi
 
 if [ "$PIP_CACHE_PURGE" = true ] ; then
-    if [ $SKIP_CONDA -eq 0 ]; then
-        # Activate conda environment
-        source activate transformers
-    fi
-
     pip cache purge
 fi

backend/python/common-env/transformers/transformers-nvidia.yml

@@ -24,10 +24,11 @@ dependencies:
   - xz=5.4.2=h5eee18b_0
   - zlib=1.2.13=h5eee18b_0
   - pip:
-      - accelerate==0.23.0
+      - accelerate==0.27.0
       - aiohttp==3.8.5
       - aiosignal==1.3.1
       - async-timeout==4.0.3
+      - auto-gptq==0.7.1
       - attrs==23.1.0
       - bark==0.1.5
       - bitsandbytes==0.43.0
@@ -69,6 +70,7 @@ dependencies:
       - nvidia-nccl-cu12==2.18.1
       - nvidia-nvjitlink-cu12==12.2.140
       - nvidia-nvtx-cu12==12.1.105
+      - optimum==1.17.1
       - packaging==23.2
       - pandas
       - peft==0.5.0
@@ -88,6 +90,7 @@ dependencies:
       - sympy==1.12
       - tokenizers
       - torch==2.1.2
+      - torchvision==0.16.2
       - torchaudio==2.1.2
       - tqdm==4.66.1
       - triton==2.1.0
@@ -95,7 +98,6 @@ dependencies:
       - tzdata==2023.3
       - urllib3==1.26.17
       - xxhash==3.4.1
-      - auto-gptq==0.6.0
       - yarl==1.9.2
       - soundfile
       - langid
@@ -114,7 +116,8 @@ dependencies:
       - sudachipy
       - sudachidict_core
       - vocos
-      - vllm==0.3.2
+      - vllm>=0.4.0
       - transformers>=4.38.2  # Updated Version
+      - transformers_stream_generator==0.0.5
       - xformers==0.0.23.post1  
 prefix: /opt/conda/envs/transformers

backend/python/common-env/transformers/transformers-rocm.yml

@@ -26,7 +26,8 @@ dependencies:
   - pip:
       - --pre
       - --extra-index-url https://download.pytorch.org/whl/nightly/
-      - accelerate==0.23.0
+      - accelerate==0.27.0
+      - auto-gptq==0.7.1
       - aiohttp==3.8.5
       - aiosignal==1.3.1
       - async-timeout==4.0.3
@@ -82,7 +83,6 @@ dependencies:
       - triton==2.1.0
       - typing-extensions==4.8.0
       - tzdata==2023.3
-      - auto-gptq==0.6.0
       - urllib3==1.26.17
       - xxhash==3.4.1
       - yarl==1.9.2
@@ -90,6 +90,7 @@ dependencies:
       - langid
       - wget
       - unidecode
+      - optimum==1.17.1
       - pyopenjtalk-prebuilt
       - pypinyin
       - inflect
@@ -103,7 +104,8 @@ dependencies:
       - sudachipy
       - sudachidict_core
       - vocos
-      - vllm==0.3.2
+      - vllm>=0.4.0
       - transformers>=4.38.2  # Updated Version
+      - transformers_stream_generator==0.0.5
       - xformers==0.0.23.post1
 prefix: /opt/conda/envs/transformers

backend/python/common-env/transformers/transformers.yml

@@ -24,15 +24,17 @@ dependencies:
   - xz=5.4.2=h5eee18b_0
   - zlib=1.2.13=h5eee18b_0
   - pip:
-      - accelerate==0.23.0
+      - accelerate==0.27.0
       - aiohttp==3.8.5
       - aiosignal==1.3.1
+      - auto-gptq==0.7.1
       - async-timeout==4.0.3
       - attrs==23.1.0
       - bark==0.1.5
       - boto3==1.28.61
       - botocore==1.31.61
       - certifi==2023.7.22
+      - coloredlogs==15.0.1
       - TTS==0.22.0
       - charset-normalizer==3.3.0
       - datasets==2.14.5
@@ -47,6 +49,7 @@ dependencies:
       - funcy==2.0
       - grpcio==1.59.0
       - huggingface-hub
+      - humanfriendly==10.0
       - idna==3.4
       - jinja2==3.1.2
       - jmespath==1.0.1
@@ -56,6 +59,10 @@ dependencies:
       - multiprocess==0.70.15
       - networkx
       - numpy==1.26.0
+      - onnx==1.15.0
+      - openvino==2024.0.0
+      - openvino-telemetry==2023.2.1
+      - optimum[openvino]==1.17.1
       - packaging==23.2
       - pandas
       - peft==0.5.0
@@ -75,12 +82,12 @@ dependencies:
       - sympy==1.12
       - tokenizers
       - torch==2.1.2
+      - torchvision==0.16.2
       - torchaudio==2.1.2
       - tqdm==4.66.1
       - triton==2.1.0
       - typing-extensions==4.8.0
       - tzdata==2023.3
-      - auto-gptq==0.6.0
       - urllib3==1.26.17
       - xxhash==3.4.1
       - yarl==1.9.2
@@ -101,7 +108,8 @@ dependencies:
       - sudachipy
       - sudachidict_core
       - vocos
-      - vllm==0.3.2
+      - vllm>=0.4.0
       - transformers>=4.38.2  # Updated Version
+      - transformers_stream_generator==0.0.5
       - xformers==0.0.23.post1  
 prefix: /opt/conda/envs/transformers

backend/python/coqui/Makefile

@@ -1,15 +1,25 @@
 .PHONY: coqui
-coqui:
+coqui: protogen
 	$(MAKE) -C ../common-env/transformers
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running coqui..."
 	bash run.sh
 	@echo "coqui run."
 
 .PHONY: test
-test:
+test: protogen
 	@echo "Testing coqui..."
 	bash test.sh
 	@echo "coqui tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/coqui/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/diffusers/Makefile

@@ -12,15 +12,25 @@ export SKIP_CONDA=1
 endif
 
 .PHONY: diffusers
-diffusers:
+diffusers: protogen
 	@echo "Installing $(CONDA_ENV_PATH)..."
 	bash install.sh $(CONDA_ENV_PATH)
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running diffusers..."
 	bash run.sh
 	@echo "Diffusers run."
 
-test:
+test: protogen
 	bash test.sh
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/diffusers/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/exllama/Makefile

@@ -1,11 +1,21 @@
 export CONDA_ENV_PATH = "exllama.yml"
 
 .PHONY: exllama
-exllama:
+exllama: protogen
 	bash install.sh ${CONDA_ENV_PATH}
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running exllama..."
 	bash run.sh
 	@echo "exllama run."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/exllama/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/exllama2/Makefile

@@ -1,10 +1,20 @@
 .PHONY: exllama2
-exllama2:
+exllama2: protogen
 	$(MAKE) -C ../common-env/transformers
 	bash install.sh
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running exllama2..."
 	bash run.sh
 	@echo "exllama2 run."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/exllama2/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/mamba/Makefile

@@ -1,16 +1,26 @@
 .PHONY: mamba
-mamba:
+mamba: protogen
 	$(MAKE) -C ../common-env/transformers
 	bash install.sh
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running mamba..."
 	bash run.sh
 	@echo "mamba run."
 
 .PHONY: test
-test:
+test: protogen
 	@echo "Testing mamba..."
 	bash test.sh
-	@echo "mamba tested."
+	@echo "mamba tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/mamba/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/parler-tts/Makefile

@@ -0,0 +1,39 @@
+export CONDA_ENV_PATH = "parler.yml"
+SKIP_CONDA?=0
+ifeq ($(BUILD_TYPE), cublas)
+export CONDA_ENV_PATH = "parler-nvidia.yml"
+endif
+
+# Intel GPU are supposed to have dependencies installed in the main python
+# environment, so we skip conda installation for SYCL builds.
+# https://github.com/intel/intel-extension-for-pytorch/issues/538
+ifneq (,$(findstring sycl,$(BUILD_TYPE)))
+export SKIP_CONDA=1
+endif
+
+.PHONY: parler-tts
+parler-tts: protogen
+	@echo "Installing $(CONDA_ENV_PATH)..."
+	bash install.sh $(CONDA_ENV_PATH)
+
+.PHONY: run
+run: protogen
+	@echo "Running transformers..."
+	bash run.sh
+	@echo "transformers run."
+
+.PHONY: test
+test: protogen
+	@echo "Testing transformers..."
+	bash test.sh
+	@echo "transformers tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/parler-tts/install.sh

@@ -0,0 +1,39 @@
+#!/bin/bash
+set -ex
+
+SKIP_CONDA=${SKIP_CONDA:-0}
+
+# Check if environment exist
+conda_env_exists(){
+    ! conda list --name "${@}" >/dev/null 2>/dev/null
+}
+
+if [ $SKIP_CONDA -eq 1 ]; then
+    echo "Skipping conda environment installation"
+else
+    export PATH=$PATH:/opt/conda/bin
+    if conda_env_exists "parler" ; then
+        echo "Creating virtual environment..."
+        conda env create --name parler --file $1
+        echo "Virtual environment created."
+    else 
+        echo "Virtual environment already exists."
+    fi
+fi
+
+if [ $SKIP_CONDA -ne 1 ]; then
+    # Activate conda environment
+    source activate parler
+    # https://github.com/descriptinc/audiotools/issues/101
+    # incompatible protobuf versions.
+    curl -L https://raw.githubusercontent.com/protocolbuffers/protobuf/main/python/google/protobuf/internal/builder.py -o $CONDA_PREFIX/lib/python3.11/site-packages/google/protobuf/internal/builder.py
+fi
+
+if [ "$PIP_CACHE_PURGE" = true ] ; then
+    if [ $SKIP_CONDA -ne 1 ]; then
+        # Activate conda environment
+        source activate parler
+    fi
+
+    pip cache purge
+fi

backend/python/parler-tts/parler-nvidia.yml

@@ -0,0 +1,48 @@
+name: parler
+channels:
+  - defaults
+dependencies:
+  - _libgcc_mutex=0.1=main
+  - _openmp_mutex=5.1=1_gnu
+  - bzip2=1.0.8=h7b6447c_0
+  - ca-certificates=2023.08.22=h06a4308_0
+  - ld_impl_linux-64=2.38=h1181459_1
+  - libffi=3.4.4=h6a678d5_0
+  - libgcc-ng=11.2.0=h1234567_1
+  - libgomp=11.2.0=h1234567_1
+  - libstdcxx-ng=11.2.0=h1234567_1
+  - libuuid=1.41.5=h5eee18b_0
+  - ncurses=6.4=h6a678d5_0
+  - openssl=3.0.11=h7f8727e_2
+  - pip=23.2.1=py311h06a4308_0
+  - python=3.11.5=h955ad1f_0
+  - readline=8.2=h5eee18b_0
+  - setuptools=68.0.0=py311h06a4308_0
+  - sqlite=3.41.2=h5eee18b_0
+  - tk=8.6.12=h1ccaba5_0
+  - tzdata=2023c=h04d1e81_0
+  - wheel=0.41.2=py311h06a4308_0
+  - xz=5.4.2=h5eee18b_0
+  - zlib=1.2.13=h5eee18b_0
+  - pip:
+      - accelerate>=0.11.0
+      - grpcio==1.59.0
+      - numpy==1.26.0
+      - nvidia-cublas-cu12==12.1.3.1
+      - nvidia-cuda-cupti-cu12==12.1.105
+      - nvidia-cuda-nvrtc-cu12==12.1.105
+      - nvidia-cuda-runtime-cu12==12.1.105
+      - nvidia-cudnn-cu12==8.9.2.26
+      - nvidia-cufft-cu12==11.0.2.54
+      - nvidia-curand-cu12==10.3.2.106
+      - nvidia-cusolver-cu12==11.4.5.107
+      - nvidia-cusparse-cu12==12.1.0.106
+      - nvidia-nccl-cu12==2.18.1
+      - nvidia-nvjitlink-cu12==12.2.140
+      - nvidia-nvtx-cu12==12.1.105
+      - torch==2.1.0
+      - transformers>=4.34.0
+      - descript-audio-codec
+      - sentencepiece
+      - git+https://github.com/huggingface/parler-tts.git@10016fb0300c0dc31a0fb70e26f3affee7b62f16
+prefix: /opt/conda/envs/diffusers

backend/python/parler-tts/parler.yml

@@ -0,0 +1,36 @@
+name: parler
+channels:
+  - defaults
+dependencies:
+  - _libgcc_mutex=0.1=main
+  - _openmp_mutex=5.1=1_gnu
+  - bzip2=1.0.8=h7b6447c_0
+  - ca-certificates=2023.08.22=h06a4308_0
+  - ld_impl_linux-64=2.38=h1181459_1
+  - libffi=3.4.4=h6a678d5_0
+  - libgcc-ng=11.2.0=h1234567_1
+  - libgomp=11.2.0=h1234567_1
+  - libstdcxx-ng=11.2.0=h1234567_1
+  - libuuid=1.41.5=h5eee18b_0
+  - ncurses=6.4=h6a678d5_0
+  - openssl=3.0.11=h7f8727e_2
+  - pip=23.2.1=py311h06a4308_0
+  - python=3.11.5=h955ad1f_0
+  - readline=8.2=h5eee18b_0
+  - setuptools=68.0.0=py311h06a4308_0
+  - sqlite=3.41.2=h5eee18b_0
+  - tk=8.6.12=h1ccaba5_0
+  - tzdata=2023c=h04d1e81_0
+  - wheel=0.41.2=py311h06a4308_0
+  - xz=5.4.2=h5eee18b_0
+  - zlib=1.2.13=h5eee18b_0
+  - pip:
+      - accelerate>=0.11.0
+      - numpy==1.26.0
+      - grpcio==1.59.0
+      - torch==2.1.0
+      - transformers>=4.34.0
+      - descript-audio-codec
+      - sentencepiece
+      - git+https://github.com/huggingface/parler-tts.git@10016fb0300c0dc31a0fb70e26f3affee7b62f16
+prefix: /opt/conda/envs/parler

backend/python/parler-tts/parler_tts_server.py

@@ -0,0 +1,125 @@
+#!/usr/bin/env python3
+"""
+Extra gRPC server for MusicgenForConditionalGeneration models.
+"""
+from concurrent import futures
+
+import argparse
+import signal
+import sys
+import os
+
+import time
+import backend_pb2
+import backend_pb2_grpc
+
+import grpc
+
+from scipy.io.wavfile import write as write_wav
+
+from parler_tts import ParlerTTSForConditionalGeneration
+from transformers import AutoTokenizer
+import soundfile as sf  
+import torch
+
+_ONE_DAY_IN_SECONDS = 60 * 60 * 24
+
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
+# Implement the BackendServicer class with the service methods
+class BackendServicer(backend_pb2_grpc.BackendServicer):
+    """
+    A gRPC servicer for the backend service.
+
+    This class implements the gRPC methods for the backend service, including Health, LoadModel, and Embedding.
+    """
+    def Health(self, request, context):
+        """
+        A gRPC method that returns the health status of the backend service.
+
+        Args:
+            request: A HealthRequest object that contains the request parameters.
+            context: A grpc.ServicerContext object that provides information about the RPC.
+
+        Returns:
+            A Reply object that contains the health status of the backend service.
+        """
+        return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
+
+    def LoadModel(self, request, context):
+        """
+        A gRPC method that loads a model into memory.
+
+        Args:
+            request: A LoadModelRequest object that contains the request parameters.
+            context: A grpc.ServicerContext object that provides information about the RPC.
+
+        Returns:
+            A Result object that contains the result of the LoadModel operation.
+        """
+        model_name = request.Model
+        device = "cuda:0" if torch.cuda.is_available() else "cpu"
+        try:
+            self.model = ParlerTTSForConditionalGeneration.from_pretrained(model_name).to(device)
+            self.tokenizer = AutoTokenizer.from_pretrained(model_name)
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+
+        return backend_pb2.Result(message="Model loaded successfully", success=True)
+
+    def TTS(self, request, context):
+        model_name = request.model
+        voice = request.voice
+        if voice == "":
+            voice = "A female speaker with a slightly low-pitched voice delivers her words quite expressively, in a very confined sounding environment with clear audio quality. She speaks very fast."
+        if model_name == "":
+            return backend_pb2.Result(success=False, message="request.model is required")
+        try:
+            device = "cuda:0" if torch.cuda.is_available() else "cpu"
+            input_ids = self.tokenizer(voice, return_tensors="pt").input_ids.to(device)
+            prompt_input_ids = self.tokenizer(request.text, return_tensors="pt").input_ids.to(device)
+           
+            generation = self.model.generate(input_ids=input_ids, prompt_input_ids=prompt_input_ids)
+            audio_arr = generation.cpu().numpy().squeeze()
+            print("[parler-tts] TTS generated!", file=sys.stderr)
+            sf.write(request.dst, audio_arr, self.model.config.sampling_rate)
+            print("[parler-tts] TTS saved to", request.dst, file=sys.stderr)
+            print("[parler-tts] TTS for", file=sys.stderr)
+            print(request, file=sys.stderr)
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+        return backend_pb2.Result(success=True)
+
+
+def serve(address):
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
+    backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
+    server.add_insecure_port(address)
+    server.start()
+    print("[parler-tts] Server started. Listening on: " + address, file=sys.stderr)
+
+    # Define the signal handler function
+    def signal_handler(sig, frame):
+        print("[parler-tts] Received termination signal. Shutting down...")
+        server.stop(0)
+        sys.exit(0)
+
+    # Set the signal handlers for SIGINT and SIGTERM
+    signal.signal(signal.SIGINT, signal_handler)
+    signal.signal(signal.SIGTERM, signal_handler)
+
+    try:
+        while True:
+            time.sleep(_ONE_DAY_IN_SECONDS)
+    except KeyboardInterrupt:
+        server.stop(0)
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Run the gRPC server.")
+    parser.add_argument(
+        "--addr", default="localhost:50051", help="The address to bind the server to."
+    )
+    args = parser.parse_args()
+    print(f"[parler-tts] startup: {args}", file=sys.stderr)
+    serve(args.addr)

backend/python/parler-tts/run.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+
+##
+## A bash script wrapper that runs the parler-tts server with conda
+
+echo "Launching gRPC server for parler-tts"
+
+export PATH=$PATH:/opt/conda/bin
+
+# Activate conda environment
+source activate parler
+
+# get the directory where the bash script is located
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
+
+python $DIR/parler_tts_server.py $@

backend/python/parler-tts/test.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+##
+## A bash script wrapper that runs the transformers server with conda
+
+# Activate conda environment
+source activate parler
+
+# get the directory where the bash script is located
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
+
+python -m unittest $DIR/test_parler.py

backend/python/parler-tts/test_parler.py

@@ -0,0 +1,81 @@
+"""
+A test script to test the gRPC service
+"""
+import unittest
+import subprocess
+import time
+import backend_pb2
+import backend_pb2_grpc
+
+import grpc
+
+
+class TestBackendServicer(unittest.TestCase):
+    """
+    TestBackendServicer is the class that tests the gRPC service
+    """
+    def setUp(self):
+        """
+        This method sets up the gRPC service by starting the server
+        """
+        self.service = subprocess.Popen(["python3", "parler_tts_server.py", "--addr", "localhost:50051"])
+        time.sleep(10)
+
+    def tearDown(self) -> None:
+        """
+        This method tears down the gRPC service by terminating the server
+        """
+        self.service.terminate()
+        self.service.wait()
+
+    def test_server_startup(self):
+        """
+        This method tests if the server starts up successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.Health(backend_pb2.HealthMessage())
+                self.assertEqual(response.message, b'OK')
+        except Exception as err:
+            print(err)
+            self.fail("Server failed to start")
+        finally:
+            self.tearDown()
+
+    def test_load_model(self):
+        """
+        This method tests if the model is loaded successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.LoadModel(backend_pb2.ModelOptions(Model="parler-tts/parler_tts_mini_v0.1"))
+                self.assertTrue(response.success)
+                self.assertEqual(response.message, "Model loaded successfully")
+        except Exception as err:
+            print(err)
+            self.fail("LoadModel service failed")
+        finally:
+            self.tearDown()
+
+    def test_tts(self):
+        """
+        This method tests if the embeddings are generated successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.LoadModel(backend_pb2.ModelOptions(Model="parler-tts/parler_tts_mini_v0.1"))
+                self.assertTrue(response.success)
+                tts_request = backend_pb2.TTSRequest(text="Hey, how are you doing today?")
+                tts_response = stub.TTS(tts_request)
+                self.assertIsNotNone(tts_response)
+        except Exception as err:
+            print(err)
+            self.fail("TTS service failed")
+        finally:
+            self.tearDown()

backend/python/petals/Makefile

@@ -1,17 +1,27 @@
 .PHONY: petals
-petals:
+petals: protogen
 	@echo "Creating virtual environment..."
 	bash install.sh "petals.yml"
 	@echo "Virtual environment created."
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running petals..."
 	bash run.sh
 	@echo "petals run."
 
 .PHONY: test
-test:
+test: protogen
 	@echo "Testing petals..."
 	bash test.sh
 	@echo "petals tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/petals/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/sentencetransformers/Makefile

@@ -1,17 +1,27 @@
 .PHONY: sentencetransformers
-sentencetransformers:
+sentencetransformers: protogen
 	$(MAKE) -C ../common-env/transformers
 
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running sentencetransformers..."
 	bash run.sh
 	@echo "sentencetransformers run."
 
 # It is not working well by using command line. It only6 works with IDE like VSCode.
 .PHONY: test
-test:
+test: protogen
 	@echo "Testing sentencetransformers..."
 	bash test.sh
 	@echo "sentencetransformers tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/sentencetransformers/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/transformers-musicgen/Makefile

@@ -1,16 +1,25 @@
-
 .PHONY: transformers-musicgen
-transformers-musicgen:
+transformers-musicgen: protogen
 	$(MAKE) -C ../common-env/transformers
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running transformers..."
 	bash run.sh
 	@echo "transformers run."
 
 .PHONY: test
-test:
+test: protogen
 	@echo "Testing transformers..."
 	bash test.sh
 	@echo "transformers tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/transformers-musicgen/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/transformers-musicgen/run.sh

@@ -8,7 +8,7 @@ echo "Launching gRPC server for transformers-musicgen"
 export PATH=$PATH:/opt/conda/bin
 
 # Activate conda environment
-source activate transformers-musicgen
+source activate transformers
 
 # get the directory where the bash script is located
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

backend/python/transformers/Makefile

@@ -1,16 +1,26 @@
 .PHONY: transformers
-transformers:
+transformers: protogen
 	$(MAKE) -C ../common-env/transformers
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running transformers..."
 	bash run.sh
 	@echo "transformers run."
 
 # It is not working well by using command line. It only6 works with IDE like VSCode.
 .PHONY: test
-test:
+test: protogen
 	@echo "Testing transformers..."
 	bash test.sh
 	@echo "transformers tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/transformers/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/transformers/transformers_server.py

@@ -8,6 +8,8 @@ import argparse
 import signal
 import sys
 import os
+from threading import Thread
+import asyncio
 
 import time
 import backend_pb2
@@ -17,13 +19,12 @@ import grpc
 import torch
 import torch.cuda
 
+
 XPU=os.environ.get("XPU", "0") == "1"
 if XPU:
-    import intel_extension_for_pytorch as ipex
-    from intel_extension_for_transformers.transformers.modeling import AutoModelForCausalLM
-    from transformers import AutoTokenizer, AutoModel, set_seed
+    from transformers import AutoTokenizer, AutoModel, set_seed, TextIteratorStreamer
 else:
-    from transformers import AutoTokenizer, AutoModel, AutoModelForCausalLM, set_seed, BitsAndBytesConfig
+    from transformers import AutoTokenizer, AutoModel, AutoModelForCausalLM, set_seed, BitsAndBytesConfig, TextIteratorStreamer
 
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
@@ -81,6 +82,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             compute=torch.bfloat16
 
         self.CUDA = request.CUDA
+        self.OV=False
 
         device_map="cpu"
 
@@ -105,23 +107,61 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                     bnb_4bit_compute_dtype = None,
                     load_in_8bit=True,                                   
                 )
-                                                   
-    
+                                               
         try:
             if request.Type == "AutoModelForCausalLM":
                 if XPU:
-                    if quantization == "xpu_4bit":
+                    import intel_extension_for_pytorch as ipex
+                    from intel_extension_for_transformers.transformers.modeling import AutoModelForCausalLM
+
+                    device_map="xpu"
+                    compute=torch.float16
+                    if request.Quantization == "xpu_4bit":
                         xpu_4bit = True
-                    self.model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=request.TrustRemoteCode,
-                                              device_map="xpu", load_in_4bit=xpu_4bit)
+                        xpu_8bit = False
+                    elif request.Quantization == "xpu_8bit":
+                        xpu_4bit = False
+                        xpu_8bit = True
+                    else:
+                        xpu_4bit = False
+                        xpu_8bit = False
+                    self.model = AutoModelForCausalLM.from_pretrained(model_name, 
+                                                                      trust_remote_code=request.TrustRemoteCode, 
+                                                                      use_safetensors=True,
+                                                                      device_map=device_map, 
+                                                                      load_in_4bit=xpu_4bit, 
+                                                                      load_in_8bit=xpu_8bit, 
+                                                                      torch_dtype=compute)
                 else:
-                    self.model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=request.TrustRemoteCode, use_safetensors=True, quantization_config=quantization, device_map=device_map, torch_dtype=compute)
+                    self.model = AutoModelForCausalLM.from_pretrained(model_name, 
+                                                                      trust_remote_code=request.TrustRemoteCode, 
+                                                                      use_safetensors=True, 
+                                                                      quantization_config=quantization, 
+                                                                      device_map=device_map, 
+                                                                      torch_dtype=compute)
+            elif request.Type == "OVModelForCausalLM":
+                from optimum.intel.openvino import OVModelForCausalLM
+                from openvino.runtime import Core
+
+                if "GPU" in Core().available_devices:
+                    device_map="GPU"
+                else:
+                    device_map="CPU"
+                self.model = OVModelForCausalLM.from_pretrained(model_name, 
+                                                                compile=True, 
+                                                                device=device_map)
+                self.OV = True
             else:
-                self.model = AutoModel.from_pretrained(model_name, trust_remote_code=request.TrustRemoteCode,  use_safetensors=True,  quantization_config=quantization, device_map=device_map, torch_dtype=compute)
+                self.model = AutoModel.from_pretrained(model_name, 
+                                                       trust_remote_code=request.TrustRemoteCode,  
+                                                       use_safetensors=True,  
+                                                       quantization_config=quantization, 
+                                                       device_map=device_map, 
+                                                       torch_dtype=compute)
             self.tokenizer = AutoTokenizer.from_pretrained(model_name, use_safetensors=True)
             self.XPU = False
 
-            if XPU:
+            if XPU and self.OV == False:
                 self.XPU = True
                 try:
                     print("Optimizing model", model_name, "to XPU.", file=sys.stderr)
@@ -130,6 +170,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
                     print("Not using XPU:", err, file=sys.stderr)
 
         except Exception as err:
+            print("Error:", err, file=sys.stderr)
             return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
         # Implement your logic here for the LoadModel service
         # Replace this with your desired response
@@ -167,7 +208,72 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         print("Embeddings:", sentence_embeddings, file=sys.stderr)
         return backend_pb2.EmbeddingResult(embeddings=sentence_embeddings[0])
 
-    def Predict(self, request, context):
+    async def _predict(self, request, context, streaming=False): 
+        set_seed(request.Seed)
+        if request.TopP == 0:
+            request.TopP = 0.9
+
+        max_tokens = 200
+        if request.Tokens > 0:
+            max_tokens = request.Tokens
+
+        inputs = self.tokenizer(request.Prompt, return_tensors="pt")
+        if self.CUDA:
+            inputs = inputs.to("cuda")
+        if XPU and self.OV == False:
+            inputs = inputs.to("xpu")
+            streaming = False
+
+        if streaming:
+            streamer=TextIteratorStreamer(self.tokenizer,
+                                        skip_prompt=True,
+                                        skip_special_tokens=True)
+            config=dict(inputs,
+                        max_new_tokens=max_tokens, 
+                        temperature=request.Temperature, 
+                        top_p=request.TopP,
+                        top_k=request.TopK, 
+                        do_sample=True,
+                        attention_mask=inputs["attention_mask"],
+                        eos_token_id=self.tokenizer.eos_token_id,
+                        pad_token_id=self.tokenizer.eos_token_id,
+                        streamer=streamer)
+            thread=Thread(target=self.model.generate, kwargs=config)
+            thread.start()
+            generated_text = ""
+            try:
+                for new_text in streamer:
+                    generated_text += new_text
+                    yield backend_pb2.Reply(message=bytes(new_text, encoding='utf-8'))
+            finally:
+                thread.join()
+        else:
+            if XPU and self.OV == False:
+                outputs = self.model.generate(inputs["input_ids"],
+                                    max_new_tokens=max_tokens, 
+                                    temperature=request.Temperature, 
+                                    top_p=request.TopP,
+                                    top_k=request.TopK, 
+                                    do_sample=True,
+                                    pad_token=self.tokenizer.eos_token_id)
+            else:
+                outputs = self.model.generate(inputs["input_ids"],
+                        max_new_tokens=max_tokens, 
+                        temperature=request.Temperature, 
+                        top_p=request.TopP,
+                        top_k=request.TopK, 
+                        do_sample=True,
+                        attention_mask=inputs["attention_mask"],
+                        eos_token_id=self.tokenizer.eos_token_id,
+                        pad_token_id=self.tokenizer.eos_token_id)
+            generated_text = self.tokenizer.batch_decode(outputs[:, inputs["input_ids"].shape[1]:], skip_special_tokens=True)[0]
+
+        if streaming:
+            return
+
+        yield backend_pb2.Reply(message=bytes(generated_text, encoding='utf-8'))
+
+    async def Predict(self, request, context):
         """
         Generates text based on the given prompt and sampling parameters.
 
@@ -178,26 +284,11 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         Returns:
             backend_pb2.Reply: The predict result.
         """
-        set_seed(request.Seed)
-        if request.TopP == 0:
-            request.TopP = 0.9
+        gen = self._predict(request, context, streaming=False)
+        res = await gen.__anext__()
+        return res
 
-        max_tokens = 200
-        if request.Tokens > 0:
-            max_tokens = request.Tokens
-
-        inputs = self.tokenizer(request.Prompt, return_tensors="pt").input_ids
-        if self.CUDA:
-            inputs = inputs.to("cuda")
-        if XPU:
-            inputs = inputs.to("xpu")
-
-        outputs = self.model.generate(inputs,max_new_tokens=max_tokens, temperature=request.Temperature, top_p=request.TopP, do_sample=True, pad_token_id=self.tokenizer.eos_token_id)
-        generated_text = self.tokenizer.batch_decode(outputs[:, inputs.shape[1]:], skip_special_tokens=True)[0]
-
-        return backend_pb2.Reply(message=bytes(generated_text, encoding='utf-8'))
-
-    def PredictStream(self, request, context):
+    async def PredictStream(self, request, context):
         """
         Generates text based on the given prompt and sampling parameters, and streams the results.
 
@@ -208,31 +299,33 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         Returns:
             backend_pb2.Result: The predict stream result.
         """
-        yield self.Predict(request, context)
+        iterations = self._predict(request, context, streaming=True)
+        try:
+            async for iteration in iterations:
+                yield iteration
+        finally:
+            await iterations.aclose()
 
-
-def serve(address):
-    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
+async def serve(address):
+    # Start asyncio gRPC server
+    server = grpc.aio.server(migration_thread_pool=futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
+    # Add the servicer to the server
     backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
+    # Bind the server to the address
     server.add_insecure_port(address)
-    server.start()
+
+    # Gracefully shutdown the server on SIGTERM or SIGINT
+    loop = asyncio.get_event_loop()
+    for sig in (signal.SIGINT, signal.SIGTERM):
+        loop.add_signal_handler(
+            sig, lambda: asyncio.ensure_future(server.stop(5))
+        )
+
+    # Start the server
+    await server.start()
     print("Server started. Listening on: " + address, file=sys.stderr)
-
-    # Define the signal handler function
-    def signal_handler(sig, frame):
-        print("Received termination signal. Shutting down...")
-        server.stop(0)
-        sys.exit(0)
-
-    # Set the signal handlers for SIGINT and SIGTERM
-    signal.signal(signal.SIGINT, signal_handler)
-    signal.signal(signal.SIGTERM, signal_handler)
-
-    try:
-        while True:
-            time.sleep(_ONE_DAY_IN_SECONDS)
-    except KeyboardInterrupt:
-        server.stop(0)
+    # Wait for the server to be terminated
+    await server.wait_for_termination()
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="Run the gRPC server.")
@@ -241,4 +334,4 @@ if __name__ == "__main__":
     )
     args = parser.parse_args()
 
-    serve(args.addr)
+    asyncio.run(serve(args.addr))

backend/python/vall-e-x/Makefile

@@ -3,18 +3,28 @@ export SKIP_CONDA=1
 endif
 
 .PHONY: ttsvalle
-ttsvalle:
+ttsvalle: protogen
 	$(MAKE) -C ../common-env/transformers
 	bash install.sh
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running ttsvalle..."
 	bash run.sh
 	@echo "ttsvalle run."
 
 .PHONY: test
-test:
+test: protogen
 	@echo "Testing valle..."
 	bash test.sh
 	@echo "valle tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/vall-e-x/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/vllm/Makefile

@@ -1,15 +1,25 @@
 .PHONY: vllm
-vllm:
+vllm: protogen
 	$(MAKE) -C ../common-env/transformers
 
 .PHONY: run
-run:
+run: protogen
 	@echo "Running vllm..."
 	bash run.sh
 	@echo "vllm run."
 
 .PHONY: test
-test:
+test: protogen
 	@echo "Testing vllm..."
 	bash test.sh
-	@echo "vllm tested."
+	@echo "vllm tested."
+
+.PHONY: protogen
+protogen: backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+backend_pb2_grpc.py backend_pb2.py:
+	python3 -m grpc_tools.protoc -I../.. --python_out=. --grpc_python_out=. backend.proto

backend/python/vllm/backend_pb2_grpc.py

@@ -1,363 +0,0 @@
-# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT!
-"""Client and server classes corresponding to protobuf-defined services."""
-import grpc
-
-import backend_pb2 as backend__pb2
-
-
-class BackendStub(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def __init__(self, channel):
-        """Constructor.
-
-        Args:
-            channel: A grpc.Channel.
-        """
-        self.Health = channel.unary_unary(
-                '/backend.Backend/Health',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Predict = channel.unary_unary(
-                '/backend.Backend/Predict',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.LoadModel = channel.unary_unary(
-                '/backend.Backend/LoadModel',
-                request_serializer=backend__pb2.ModelOptions.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.PredictStream = channel.unary_stream(
-                '/backend.Backend/PredictStream',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.Reply.FromString,
-                )
-        self.Embedding = channel.unary_unary(
-                '/backend.Backend/Embedding',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.EmbeddingResult.FromString,
-                )
-        self.GenerateImage = channel.unary_unary(
-                '/backend.Backend/GenerateImage',
-                request_serializer=backend__pb2.GenerateImageRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.AudioTranscription = channel.unary_unary(
-                '/backend.Backend/AudioTranscription',
-                request_serializer=backend__pb2.TranscriptRequest.SerializeToString,
-                response_deserializer=backend__pb2.TranscriptResult.FromString,
-                )
-        self.TTS = channel.unary_unary(
-                '/backend.Backend/TTS',
-                request_serializer=backend__pb2.TTSRequest.SerializeToString,
-                response_deserializer=backend__pb2.Result.FromString,
-                )
-        self.TokenizeString = channel.unary_unary(
-                '/backend.Backend/TokenizeString',
-                request_serializer=backend__pb2.PredictOptions.SerializeToString,
-                response_deserializer=backend__pb2.TokenizationResponse.FromString,
-                )
-        self.Status = channel.unary_unary(
-                '/backend.Backend/Status',
-                request_serializer=backend__pb2.HealthMessage.SerializeToString,
-                response_deserializer=backend__pb2.StatusResponse.FromString,
-                )
-
-
-class BackendServicer(object):
-    """Missing associated documentation comment in .proto file."""
-
-    def Health(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Predict(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def LoadModel(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def PredictStream(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Embedding(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def GenerateImage(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def AudioTranscription(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TTS(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def TokenizeString(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-    def Status(self, request, context):
-        """Missing associated documentation comment in .proto file."""
-        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
-        context.set_details('Method not implemented!')
-        raise NotImplementedError('Method not implemented!')
-
-
-def add_BackendServicer_to_server(servicer, server):
-    rpc_method_handlers = {
-            'Health': grpc.unary_unary_rpc_method_handler(
-                    servicer.Health,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Predict': grpc.unary_unary_rpc_method_handler(
-                    servicer.Predict,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'LoadModel': grpc.unary_unary_rpc_method_handler(
-                    servicer.LoadModel,
-                    request_deserializer=backend__pb2.ModelOptions.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'PredictStream': grpc.unary_stream_rpc_method_handler(
-                    servicer.PredictStream,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.Reply.SerializeToString,
-            ),
-            'Embedding': grpc.unary_unary_rpc_method_handler(
-                    servicer.Embedding,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.EmbeddingResult.SerializeToString,
-            ),
-            'GenerateImage': grpc.unary_unary_rpc_method_handler(
-                    servicer.GenerateImage,
-                    request_deserializer=backend__pb2.GenerateImageRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'AudioTranscription': grpc.unary_unary_rpc_method_handler(
-                    servicer.AudioTranscription,
-                    request_deserializer=backend__pb2.TranscriptRequest.FromString,
-                    response_serializer=backend__pb2.TranscriptResult.SerializeToString,
-            ),
-            'TTS': grpc.unary_unary_rpc_method_handler(
-                    servicer.TTS,
-                    request_deserializer=backend__pb2.TTSRequest.FromString,
-                    response_serializer=backend__pb2.Result.SerializeToString,
-            ),
-            'TokenizeString': grpc.unary_unary_rpc_method_handler(
-                    servicer.TokenizeString,
-                    request_deserializer=backend__pb2.PredictOptions.FromString,
-                    response_serializer=backend__pb2.TokenizationResponse.SerializeToString,
-            ),
-            'Status': grpc.unary_unary_rpc_method_handler(
-                    servicer.Status,
-                    request_deserializer=backend__pb2.HealthMessage.FromString,
-                    response_serializer=backend__pb2.StatusResponse.SerializeToString,
-            ),
-    }
-    generic_handler = grpc.method_handlers_generic_handler(
-            'backend.Backend', rpc_method_handlers)
-    server.add_generic_rpc_handlers((generic_handler,))
-
-
- # This class is part of an EXPERIMENTAL API.
-class Backend(object):
-    """Missing associated documentation comment in .proto file."""
-
-    @staticmethod
-    def Health(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Health',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Predict(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Predict',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def LoadModel(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/LoadModel',
-            backend__pb2.ModelOptions.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def PredictStream(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_stream(request, target, '/backend.Backend/PredictStream',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.Reply.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Embedding(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Embedding',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.EmbeddingResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def GenerateImage(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/GenerateImage',
-            backend__pb2.GenerateImageRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def AudioTranscription(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/AudioTranscription',
-            backend__pb2.TranscriptRequest.SerializeToString,
-            backend__pb2.TranscriptResult.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TTS(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TTS',
-            backend__pb2.TTSRequest.SerializeToString,
-            backend__pb2.Result.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def TokenizeString(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/TokenizeString',
-            backend__pb2.PredictOptions.SerializeToString,
-            backend__pb2.TokenizationResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
-
-    @staticmethod
-    def Status(request,
-            target,
-            options=(),
-            channel_credentials=None,
-            call_credentials=None,
-            insecure=False,
-            compression=None,
-            wait_for_ready=None,
-            timeout=None,
-            metadata=None):
-        return grpc.experimental.unary_unary(request, target, '/backend.Backend/Status',
-            backend__pb2.HealthMessage.SerializeToString,
-            backend__pb2.StatusResponse.FromString,
-            options, channel_credentials,
-            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

backend/python/vllm/backend_vllm.py

@@ -14,6 +14,7 @@ from vllm.engine.arg_utils import AsyncEngineArgs
 from vllm.engine.async_llm_engine import AsyncLLMEngine
 from vllm.sampling_params import SamplingParams
 from vllm.utils import random_uuid
+from vllm.transformers_utils.tokenizer import get_tokenizer
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
@@ -71,7 +72,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         """
         return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
 
-    def LoadModel(self, request, context):
+    async def LoadModel(self, request, context):
         """
         Loads a language model.
 
@@ -103,6 +104,18 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             self.llm = AsyncLLMEngine.from_engine_args(engine_args)
         except Exception as err:
             return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+
+        try:
+           engine_model_config = await self.llm.get_model_config()
+           self.tokenizer = get_tokenizer(
+               engine_model_config.tokenizer,
+               tokenizer_mode=engine_model_config.tokenizer_mode,
+               trust_remote_code=engine_model_config.trust_remote_code,
+               truncation_side="left",
+           )
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+
         return backend_pb2.Result(message="Model loaded successfully", success=True)
 
     async def Predict(self, request, context):
@@ -161,9 +174,15 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         if request.Seed != 0:
             sampling_params.seed = request.Seed
 
+        prompt = request.Prompt
+        
+        # If tokenizer template is enabled and messages are provided instead of prompt apply the tokenizer template
+        if not request.Prompt and request.UseTokenizerTemplate and request.Messages:
+            prompt = self.tokenizer.apply_chat_template(request.Messages, tokenize=False, add_generation_prompt=True)
+
         # Generate text
         request_id = random_uuid()
-        outputs = self.llm.generate(request.Prompt, sampling_params, request_id)
+        outputs = self.llm.generate(prompt, sampling_params, request_id)
 
         # Stream the results
         generated_text = ""

core/backend/embeddings.go

@@ -2,14 +2,100 @@ package backend
 
 import (
 	"fmt"
+	"time"
 
 	"github.com/go-skynet/LocalAI/core/config"
+	"github.com/go-skynet/LocalAI/core/schema"
+	"github.com/google/uuid"
 
+	"github.com/go-skynet/LocalAI/pkg/concurrency"
 	"github.com/go-skynet/LocalAI/pkg/grpc"
-	model "github.com/go-skynet/LocalAI/pkg/model"
+	"github.com/go-skynet/LocalAI/pkg/model"
 )
 
-func ModelEmbedding(s string, tokens []int, loader *model.ModelLoader, backendConfig config.BackendConfig, appConfig *config.ApplicationConfig) (func() ([]float32, error), error) {
+type EmbeddingsBackendService struct {
+	ml        *model.ModelLoader
+	bcl       *config.BackendConfigLoader
+	appConfig *config.ApplicationConfig
+}
+
+func NewEmbeddingsBackendService(ml *model.ModelLoader, bcl *config.BackendConfigLoader, appConfig *config.ApplicationConfig) *EmbeddingsBackendService {
+	return &EmbeddingsBackendService{
+		ml:        ml,
+		bcl:       bcl,
+		appConfig: appConfig,
+	}
+}
+
+func (ebs *EmbeddingsBackendService) Embeddings(request *schema.OpenAIRequest) <-chan concurrency.ErrorOr[*schema.OpenAIResponse] {
+
+	resultChannel := make(chan concurrency.ErrorOr[*schema.OpenAIResponse])
+	go func(request *schema.OpenAIRequest) {
+		if request.Model == "" {
+			request.Model = model.StableDiffusionBackend
+		}
+
+		bc, request, err := ebs.bcl.LoadBackendConfigForModelAndOpenAIRequest(request.Model, request, ebs.appConfig)
+		if err != nil {
+			resultChannel <- concurrency.ErrorOr[*schema.OpenAIResponse]{Error: err}
+			close(resultChannel)
+			return
+		}
+
+		items := []schema.Item{}
+
+		for i, s := range bc.InputToken {
+			// get the model function to call for the result
+			embedFn, err := modelEmbedding("", s, ebs.ml, bc, ebs.appConfig)
+			if err != nil {
+				resultChannel <- concurrency.ErrorOr[*schema.OpenAIResponse]{Error: err}
+				close(resultChannel)
+				return
+			}
+
+			embeddings, err := embedFn()
+			if err != nil {
+				resultChannel <- concurrency.ErrorOr[*schema.OpenAIResponse]{Error: err}
+				close(resultChannel)
+				return
+			}
+			items = append(items, schema.Item{Embedding: embeddings, Index: i, Object: "embedding"})
+		}
+
+		for i, s := range bc.InputStrings {
+			// get the model function to call for the result
+			embedFn, err := modelEmbedding(s, []int{}, ebs.ml, bc, ebs.appConfig)
+			if err != nil {
+				resultChannel <- concurrency.ErrorOr[*schema.OpenAIResponse]{Error: err}
+				close(resultChannel)
+				return
+			}
+
+			embeddings, err := embedFn()
+			if err != nil {
+				resultChannel <- concurrency.ErrorOr[*schema.OpenAIResponse]{Error: err}
+				close(resultChannel)
+				return
+			}
+			items = append(items, schema.Item{Embedding: embeddings, Index: i, Object: "embedding"})
+		}
+
+		id := uuid.New().String()
+		created := int(time.Now().Unix())
+		resp := &schema.OpenAIResponse{
+			ID:      id,
+			Created: created,
+			Model:   request.Model, // we have to return what the user sent here, due to OpenAI spec.
+			Data:    items,
+			Object:  "list",
+		}
+		resultChannel <- concurrency.ErrorOr[*schema.OpenAIResponse]{Value: resp}
+		close(resultChannel)
+	}(request)
+	return resultChannel
+}
+
+func modelEmbedding(s string, tokens []int, loader *model.ModelLoader, backendConfig *config.BackendConfig, appConfig *config.ApplicationConfig) (func() ([]float32, error), error) {
 	modelFile := backendConfig.Model
 
 	grpcOpts := gRPCModelOpts(backendConfig)