fix(autogptq): do not use_triton with qwen-vl (#1985)

* Enhance autogptq backend to support VL models

* update dependencies for autogptq

* remove redundant auto-gptq dependency

* Convert base64 to image_url for Qwen-VL model

* implemented model inference for qwen-vl

* remove user prompt from generated answer

* fixed write image error

* fixed use_triton issue when loading Qwen-VL model

---------

Co-authored-by: Binghua Wu <bingwu@estee.com>

.dockerignore

@@ -1,16 +1,6 @@
 .idea
-.github
-.vscode
 models
 examples/chatbot-ui/models
 examples/rwkv/models
 examples/**/models
-Dockerfile*
-__pycache__
-
-# SonarQube
-.scannerwork
-
-# backend virtual environments
-**/venv
-backend/python/**/source
+Dockerfile*

.env

@@ -1,33 +1,33 @@
 ## Set number of threads.
 ## Note: prefer the number of physical cores. Overbooking the CPU degrades performance notably.
-# LOCALAI_THREADS=14
+# THREADS=14
 
 ## Specify a different bind address (defaults to ":8080")
-# LOCALAI_ADDRESS=127.0.0.1:8080
+# ADDRESS=127.0.0.1:8080
 
 ## Default models context size
-# LOCALAI_CONTEXT_SIZE=512
+# CONTEXT_SIZE=512
 #
 ## Define galleries.
 ## models will to install will be visible in `/models/available`
-# LOCALAI_GALLERIES=[{"name":"localai", "url":"github:mudler/LocalAI/gallery/index.yaml@master"}]
+# GALLERIES=[{"name":"model-gallery", "url":"github:go-skynet/model-gallery/index.yaml"}]
 
 ## CORS settings
-# LOCALAI_CORS=true
-# LOCALAI_CORS_ALLOW_ORIGINS=*
+# CORS=true
+# CORS_ALLOW_ORIGINS=*
 
 ## Default path for models
 #
-# LOCALAI_MODELS_PATH=/models
+# MODELS_PATH=/models
 
 ## Enable debug mode
-# LOCALAI_LOG_LEVEL=debug
+# DEBUG=true
 
 ## Disables COMPEL (Diffusers)
 # COMPEL=0
 
 ## Enable/Disable single backend (useful if only one GPU is available)
-# LOCALAI_SINGLE_ACTIVE_BACKEND=true
+# 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
-# LOCALAI_IMAGE_PATH=/tmp/generated/images
+# IMAGE_PATH=/tmp
 
 ## Specify a default upload limit in MB (whisper)
-# LOCALAI_UPLOAD_LIMIT=15
+# UPLOAD_LIMIT
 
 ## List of external GRPC backends (note on the container image this variable is already set to use extra backends available in extra/)
-# LOCALAI_EXTERNAL_GRPC_BACKENDS=my-backend:127.0.0.1:9000,my-backend2:/usr/bin/backend.py
+# 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
-# LOCALAI_PARALLEL_REQUESTS=true
+# PARALLEL_REQUESTS=true
 
 ### Watchdog settings
 ###
 # Enables watchdog to kill backends that are inactive for too much time
-# LOCALAI_WATCHDOG_IDLE=true
-#
-# Time in duration format (e.g. 1h30m) after which a backend is considered idle
-# LOCALAI_WATCHDOG_IDLE_TIMEOUT=5m
+# WATCHDOG_IDLE=true
 #
 # Enables watchdog to kill backends that are busy for too much time
-# LOCALAI_WATCHDOG_BUSY=true
+# WATCHDOG_BUSY=true
+#
+# Time in duration format (e.g. 1h30m) after which a backend is considered idle
+# WATCHDOG_IDLE_TIMEOUT=5m
 #
 # Time in duration format (e.g. 1h30m) after which a backend is considered busy
-# LOCALAI_WATCHDOG_BUSY_TIMEOUT=5m
+# WATCHDOG_BUSY_TIMEOUT=5m

.github/bump_docs.sh

@@ -2,6 +2,6 @@
 set -xe
 REPO=$1
 
-LATEST_TAG=$(curl -s "https://api.github.com/repos/$REPO/releases/latest" | jq -r '.tag_name')
+LATEST_TAG=$(curl -s "https://api.github.com/repos/$REPO/releases/latest" | jq -r '.name')
 
 cat <<< $(jq ".version = \"$LATEST_TAG\"" docs/data/version.json) > docs/data/version.json

.github/checksum_checker.sh

@@ -1,111 +0,0 @@
-#!/bin/bash
-# This scripts needs yq and huggingface_hub to be installed
-# to install hugingface_hub run pip install huggingface_hub
-
-# Path to the input YAML file
-input_yaml=$1
-
-# Function to download file and check checksum using Python
-function check_and_update_checksum() {
-    model_name="$1"
-    file_name="$2"
-    uri="$3"
-    old_checksum="$4"
-    idx="$5"
-
-    # Download the file and calculate new checksum using Python
-    new_checksum=$(python3 -c "
-import hashlib
-from huggingface_hub import hf_hub_download
-import requests
-import sys
-import os
-
-uri = '$uri'
-file_name = uri.split('/')[-1]
-
-# Function to parse the URI and determine download method
-# Function to parse the URI and determine download method
-def parse_uri(uri):
-    if uri.startswith('huggingface://'):
-        repo_id = uri.split('://')[1]
-        return 'huggingface', repo_id.rsplit('/', 1)[0]
-    elif 'huggingface.co' in uri:
-        parts = uri.split('/resolve/')
-        if len(parts) > 1:
-            repo_path = parts[0].split('https://huggingface.co/')[-1]
-            return 'huggingface', repo_path
-    return 'direct', uri
-
-def calculate_sha256(file_path):
-    sha256_hash = hashlib.sha256()
-    with open(file_path, 'rb') as f:
-        for byte_block in iter(lambda: f.read(4096), b''):
-            sha256_hash.update(byte_block)
-    return sha256_hash.hexdigest()
-
-download_type, repo_id_or_url = parse_uri(uri)
-
-# Decide download method based on URI type
-if download_type == 'huggingface':
-    try:
-        file_path = hf_hub_download(repo_id=repo_id_or_url, filename=file_name)
-    except Exception as e:
-        print(f'Error from Hugging Face Hub: {str(e)}', file=sys.stderr)
-        sys.exit(2)
-else:
-    response = requests.get(repo_id_or_url)
-    if response.status_code == 200:
-        with open(file_name, 'wb') as f:
-            f.write(response.content)
-        file_path = file_name
-    elif response.status_code == 404:
-        print(f'File not found: {response.status_code}', file=sys.stderr)
-        sys.exit(2)
-    else:
-        print(f'Error downloading file: {response.status_code}', file=sys.stderr)
-        sys.exit(1)
-
-print(calculate_sha256(file_path))
-# Clean up the downloaded file
-os.remove(file_path)
-")
-
-    if [[ "$new_checksum" == "" ]]; then
-        echo "Error calculating checksum for $file_name. Skipping..."
-        return
-    fi
-
-    echo "Checksum for $file_name: $new_checksum"
-
-    # Compare and update the YAML file if checksums do not match
-    result=$?
-    if [[ $result -eq 2 ]]; then
-        echo "File not found, deleting entry for $file_name..."
-        # yq eval -i "del(.[$idx].files[] | select(.filename == \"$file_name\"))" "$input_yaml"
-    elif [[ "$old_checksum" != "$new_checksum" ]]; then
-        echo "Checksum mismatch for $file_name. Updating..."
-        yq eval -i "del(.[$idx].files[] | select(.filename == \"$file_name\").sha256)" "$input_yaml"
-        yq eval -i "(.[$idx].files[] | select(.filename == \"$file_name\")).sha256 = \"$new_checksum\"" "$input_yaml"
-    elif [[ $result -ne 0 ]]; then
-        echo "Error downloading file $file_name. Skipping..."
-    else
-        echo "Checksum match for $file_name. No update needed."
-    fi
-}
-
-# Read the YAML and process each file
-len=$(yq eval '. | length' "$input_yaml")
-for ((i=0; i<$len; i++))
-do
-    name=$(yq eval ".[$i].name" "$input_yaml")
-    files_len=$(yq eval ".[$i].files | length" "$input_yaml")
-    for ((j=0; j<$files_len; j++))
-    do
-        filename=$(yq eval ".[$i].files[$j].filename" "$input_yaml")
-        uri=$(yq eval ".[$i].files[$j].uri" "$input_yaml")
-        checksum=$(yq eval ".[$i].files[$j].sha256" "$input_yaml")
-        echo "Checking model $name, file $filename. URI = $uri, Checksum = $checksum"
-        check_and_update_checksum "$name" "$filename" "$uri" "$checksum" "$i"
-    done
-done

.github/dependabot.yml

@@ -1,25 +0,0 @@
-# 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

@@ -8,11 +8,6 @@ kind/documentation:
   - changed-files:
     - any-glob-to-any-file: '*.md'
 
-area/ai-model:
-- any:
-  - changed-files:
-    - any-glob-to-any-file: 'gallery/*'
-
 examples:
 - any:
   - changed-files:
@@ -21,4 +16,4 @@ examples:
 ci:
 - any:
   - changed-files:
-    - any-glob-to-any-file: '.github/*'
+    - 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@v6
+        uses: peter-evans/create-pull-request@v5
         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@v6
+        uses: peter-evans/create-pull-request@v5
         with:
           token: ${{ secrets.UPDATE_BOT_TOKEN }}
           push-to-fork: ci-forks/LocalAI

.github/workflows/checksum_checker.yaml

@@ -1,47 +0,0 @@
-name: Check if checksums are up-to-date
-on:
-  schedule:
-    - cron: 0 20 * * *
-  workflow_dispatch:
-jobs:
-  checksum_check:
-    runs-on: arc-runner-set
-    steps:
-      - name: Force Install GIT latest
-        run: |
-          sudo apt-get update \
-          && sudo apt-get install -y software-properties-common \
-          && sudo apt-get update \
-          && sudo add-apt-repository -y ppa:git-core/ppa \
-          && sudo apt-get update \
-          && sudo apt-get install -y git
-      - uses: actions/checkout@v4
-      - name: Install dependencies
-        run: |
-          sudo apt-get update
-          sudo apt-get install -y pip wget
-          sudo pip install --upgrade pip 
-          pip install huggingface_hub
-      - name: 'Setup yq'
-        uses: dcarbone/install-yq-action@v1.1.1
-        with:
-          version: 'v4.43.1'
-          download-compressed: true
-          force: true
-
-      - name: Checksum checker 🔧
-        run: |
-          export HF_HOME=/hf_cache
-          sudo mkdir /hf_cache
-          sudo chmod 777 /hf_cache
-          bash .github/checksum_checker.sh gallery/index.yaml
-      - name: Create Pull Request
-        uses: peter-evans/create-pull-request@v6
-        with:
-          token: ${{ secrets.UPDATE_BOT_TOKEN }}
-          push-to-fork: ci-forks/LocalAI
-          commit-message: ':arrow_up: Checksum updates in gallery/index.yaml'
-          title: 'models(gallery): :arrow_up: update checksum'
-          branch: "update/checksum"
-          body: Updating checksums in gallery/index.yaml
-          signoff: true

.github/workflows/dependabot_auto.yml

@@ -1,43 +0,0 @@
-name: Dependabot auto-merge
-on:
-- pull_request_target
-
-permissions:
-  contents: write
-  pull-requests: write
-  packages: read
-
-jobs:
-  dependabot:
-    runs-on: ubuntu-latest
-    if: ${{ github.actor == 'dependabot[bot]' }}
-    steps:
-      - name: Dependabot metadata
-        id: metadata
-        uses: dependabot/fetch-metadata@v2.1.0
-        with:
-          github-token: "${{ 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/generate_grpc_cache.yaml

@@ -1,94 +0,0 @@
-name: 'generate and publish GRPC docker caches'
-
-on:
-  workflow_dispatch:
-  push:
-    branches:
-      - master
-
-concurrency:
-  group: grpc-cache-${{ github.head_ref || github.ref }}-${{ github.repository }}
-  cancel-in-progress: true
-
-jobs:
-  generate_caches:
-    strategy:
-      matrix:
-        include:
-          - grpc-base-image: ubuntu:22.04
-            runs-on: 'ubuntu-latest'
-            platforms: 'linux/amd64'
-    runs-on: ${{matrix.runs-on}}
-    steps:
-      - name: Release space from worker
-        if: matrix.runs-on == 'ubuntu-latest'
-        run: |
-          echo "Listing top largest packages"
-          pkgs=$(dpkg-query -Wf '${Installed-Size}\t${Package}\t${Status}\n' | awk '$NF == "installed"{print $1 "\t" $2}' | sort -nr)
-          head -n 30 <<< "${pkgs}"
-          echo
-          df -h
-          echo
-          sudo apt-get remove -y '^llvm-.*|^libllvm.*' || true
-          sudo apt-get remove --auto-remove android-sdk-platform-tools || true
-          sudo apt-get purge --auto-remove android-sdk-platform-tools || true
-          sudo rm -rf /usr/local/lib/android
-          sudo apt-get remove -y '^dotnet-.*|^aspnetcore-.*' || true
-          sudo rm -rf /usr/share/dotnet
-          sudo apt-get remove -y '^mono-.*' || true
-          sudo apt-get remove -y '^ghc-.*' || true
-          sudo apt-get remove -y '.*jdk.*|.*jre.*' || true
-          sudo apt-get remove -y 'php.*' || true
-          sudo apt-get remove -y hhvm powershell firefox monodoc-manual msbuild || true
-          sudo apt-get remove -y '^google-.*' || true
-          sudo apt-get remove -y azure-cli || true
-          sudo apt-get remove -y '^mongo.*-.*|^postgresql-.*|^mysql-.*|^mssql-.*' || true
-          sudo apt-get remove -y '^gfortran-.*' || true
-          sudo apt-get remove -y microsoft-edge-stable || true
-          sudo apt-get remove -y firefox || true
-          sudo apt-get remove -y powershell || true
-          sudo apt-get remove -y r-base-core || true
-          sudo apt-get autoremove -y
-          sudo apt-get clean
-          echo
-          echo "Listing top largest packages"
-          pkgs=$(dpkg-query -Wf '${Installed-Size}\t${Package}\t${Status}\n' | awk '$NF == "installed"{print $1 "\t" $2}' | sort -nr)
-          head -n 30 <<< "${pkgs}"
-          echo
-          sudo rm -rfv build || true
-          sudo rm -rf /usr/share/dotnet || true
-          sudo rm -rf /opt/ghc || true
-          sudo rm -rf "/usr/local/share/boost" || true
-          sudo rm -rf "$AGENT_TOOLSDIRECTORY" || true
-          df -h
-
-      - name: Set up QEMU
-        uses: docker/setup-qemu-action@master
-        with:
-          platforms: all
-
-      - name: Set up Docker Buildx
-        id: buildx
-        uses: docker/setup-buildx-action@master
-
-      - name: Checkout
-        uses: actions/checkout@v4
-
-      - name: Cache GRPC
-        uses: docker/build-push-action@v5
-        with:
-          builder: ${{ steps.buildx.outputs.name }}
-          # The build-args MUST be an EXACT match between the image cache and other workflow steps that want to use that cache.
-          # This means that even the MAKEFLAGS have to be an EXACT match.
-          # If the build-args are not an EXACT match, it will result in a cache miss, which will require GRPC to be built from scratch.
-          build-args: |
-            GRPC_BASE_IMAGE=${{ matrix.grpc-base-image }}
-            GRPC_MAKEFLAGS=--jobs=4 --output-sync=target
-            GRPC_VERSION=v1.63.0
-          context: .
-          file: ./Dockerfile
-          cache-to: type=gha,ignore-error=true
-          cache-from: type=gha
-          target: grpc
-          platforms: ${{ matrix.platforms }}
-          push: false

.github/workflows/image-pr.yml

@@ -22,7 +22,6 @@ jobs:
       platforms: ${{ matrix.platforms }}
       runs-on: ${{ matrix.runs-on }}
       base-image: ${{ matrix.base-image }}
-      grpc-base-image: ${{ matrix.grpc-base-image }}
       makeflags: ${{ matrix.makeflags }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
@@ -61,15 +60,13 @@ jobs:
             tag-suffix: '-hipblas'
             ffmpeg: 'false'
             image-type: 'extras'
-            base-image: "rocm/dev-ubuntu-22.04:6.1"
-            grpc-base-image: "ubuntu:22.04"
+            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'
-            base-image: "intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
             tag-suffix: 'sycl-f16-ffmpeg'
             ffmpeg: 'true'
             image-type: 'extras'
@@ -88,7 +85,6 @@ jobs:
       platforms: ${{ matrix.platforms }}
       runs-on: ${{ matrix.runs-on }}
       base-image: ${{ matrix.base-image }}
-      grpc-base-image: ${{ matrix.grpc-base-image }}
       makeflags: ${{ matrix.makeflags }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
@@ -106,12 +102,11 @@ jobs:
             image-type: 'core'
             runs-on: 'ubuntu-latest'
             base-image: "ubuntu:22.04"
-            makeflags: "--jobs=4 --output-sync=target"
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'sycl_f16'
             platforms: 'linux/amd64'
             tag-latest: 'false'
-            base-image: "intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
             tag-suffix: 'sycl-f16-ffmpeg-core'
             ffmpeg: 'true'
             image-type: 'core'
@@ -127,4 +122,4 @@ jobs:
             image-type: 'core'
             runs-on: 'ubuntu-latest'
             base-image: "ubuntu:22.04"
-            makeflags: "--jobs=4 --output-sync=target"
+            makeflags: "--jobs=5 --output-sync=target"

.github/workflows/image.yml

@@ -26,7 +26,6 @@ jobs:
       platforms: ${{ matrix.platforms }}
       runs-on: ${{ matrix.runs-on }}
       base-image: ${{ matrix.base-image }}
-      grpc-base-image: ${{ matrix.grpc-base-image }}
       aio: ${{ matrix.aio }}
       makeflags: ${{ matrix.makeflags }}
       latest-image: ${{ matrix.latest-image }}
@@ -129,8 +128,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'extras'
             aio: "-aio-gpu-hipblas"
-            base-image: "rocm/dev-ubuntu-22.04:6.1"
-            grpc-base-image: "ubuntu:22.04"
+            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'
@@ -141,15 +139,13 @@ jobs:
             tag-suffix: '-hipblas'
             ffmpeg: 'false'
             image-type: 'extras'
-            base-image: "rocm/dev-ubuntu-22.04:6.1"
-            grpc-base-image: "ubuntu:22.04"
+            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'
-            base-image: "intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
             tag-suffix: '-sycl-f16-ffmpeg'
             ffmpeg: 'true'
             image-type: 'extras'
@@ -161,8 +157,7 @@ jobs:
           - build-type: 'sycl_f32'
             platforms: 'linux/amd64'
             tag-latest: 'auto'
-            base-image: "intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
             tag-suffix: '-sycl-f32-ffmpeg'
             ffmpeg: 'true'
             image-type: 'extras'
@@ -175,8 +170,7 @@ jobs:
           - build-type: 'sycl_f16'
             platforms: 'linux/amd64'
             tag-latest: 'false'
-            base-image: "intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
             tag-suffix: '-sycl-f16-core'
             ffmpeg: 'false'
             image-type: 'core'
@@ -185,8 +179,7 @@ jobs:
           - build-type: 'sycl_f32'
             platforms: 'linux/amd64'
             tag-latest: 'false'
-            base-image: "intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
             tag-suffix: '-sycl-f32-core'
             ffmpeg: 'false'
             image-type: 'core'
@@ -195,8 +188,7 @@ jobs:
           - build-type: 'sycl_f16'
             platforms: 'linux/amd64'
             tag-latest: 'false'
-            base-image: "intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
             tag-suffix: '-sycl-f16-ffmpeg-core'
             ffmpeg: 'true'
             image-type: 'core'
@@ -205,8 +197,7 @@ jobs:
           - build-type: 'sycl_f32'
             platforms: 'linux/amd64'
             tag-latest: 'false'
-            base-image: "intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
             tag-suffix: '-sycl-f32-ffmpeg-core'
             ffmpeg: 'true'
             image-type: 'core'
@@ -218,8 +209,7 @@ jobs:
             tag-suffix: '-hipblas-ffmpeg-core'
             ffmpeg: 'true'
             image-type: 'core'
-            base-image: "rocm/dev-ubuntu-22.04:6.1"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "rocm/dev-ubuntu-22.04:6.0-complete"
             runs-on: 'arc-runner-set'
             makeflags: "--jobs=3 --output-sync=target"
           - build-type: 'hipblas'
@@ -228,8 +218,7 @@ jobs:
             tag-suffix: '-hipblas-core'
             ffmpeg: 'false'
             image-type: 'core'
-            base-image: "rocm/dev-ubuntu-22.04:6.1"
-            grpc-base-image: "ubuntu:22.04"
+            base-image: "rocm/dev-ubuntu-22.04:6.0-complete"
             runs-on: 'arc-runner-set'
             makeflags: "--jobs=3 --output-sync=target"
   
@@ -247,7 +236,6 @@ jobs:
       runs-on: ${{ matrix.runs-on }}
       aio: ${{ matrix.aio }}
       base-image: ${{ matrix.base-image }}
-      grpc-base-image: ${{ matrix.grpc-base-image }}
       makeflags: ${{ matrix.makeflags }}
       latest-image: ${{ matrix.latest-image }}
       latest-image-aio: ${{ matrix.latest-image-aio }}
@@ -270,7 +258,7 @@ jobs:
             aio: "-aio-cpu"
             latest-image: 'latest-cpu'
             latest-image-aio: 'latest-aio-cpu'
-            makeflags: "--jobs=4 --output-sync=target"
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "11"
             cuda-minor-version: "7"
@@ -281,7 +269,7 @@ jobs:
             image-type: 'core'
             base-image: "ubuntu:22.04"
             runs-on: 'ubuntu-latest'
-            makeflags: "--jobs=4 --output-sync=target"
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -292,7 +280,7 @@ jobs:
             image-type: 'core'
             base-image: "ubuntu:22.04"
             runs-on: 'ubuntu-latest'
-            makeflags: "--jobs=4 --output-sync=target"
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "11"
             cuda-minor-version: "7"
@@ -303,7 +291,7 @@ jobs:
             image-type: 'core'
             runs-on: 'ubuntu-latest'
             base-image: "ubuntu:22.04"
-            makeflags: "--jobs=4 --output-sync=target"
+            makeflags: "--jobs=5 --output-sync=target"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -314,4 +302,4 @@ jobs:
             image-type: 'core'
             runs-on: 'ubuntu-latest'
             base-image: "ubuntu:22.04"
-            makeflags: "--jobs=4 --output-sync=target"
+            makeflags: "--jobs=5 --output-sync=target"

.github/workflows/image_build.yml

@@ -6,10 +6,6 @@ on:
     inputs:
       base-image:
         description: 'Base image'
-        required: true
-        type: string
-      grpc-base-image:
-        description: 'GRPC Base image, must be a compatible image with base-image'
         required: false
         default: ''
         type: string
@@ -61,7 +57,7 @@ on:
       makeflags:
         description: 'Make Flags'
         required: false
-        default: '--jobs=4 --output-sync=target'
+        default: '--jobs=3 --output-sync=target'
         type: string
       aio:
         description: 'AIO Image Name'
@@ -201,14 +197,29 @@ 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:
           builder: ${{ steps.buildx.outputs.name }}
-          # The build-args MUST be an EXACT match between the image cache and other workflow steps that want to use that cache.
-          # This means that even the MAKEFLAGS have to be an EXACT match.
-          # If the build-args are not an EXACT match, it will result in a cache miss, which will require GRPC to be built from scratch.
-          # This is why some build args like GRPC_VERSION and MAKEFLAGS are hardcoded
           build-args: |
             BUILD_TYPE=${{ inputs.build-type }}
             CUDA_MAJOR_VERSION=${{ inputs.cuda-major-version }}
@@ -216,9 +227,6 @@ jobs:
             FFMPEG=${{ inputs.ffmpeg }}
             IMAGE_TYPE=${{ inputs.image-type }}
             BASE_IMAGE=${{ inputs.base-image }}
-            GRPC_BASE_IMAGE=${{ inputs.grpc-base-image || inputs.base-image }}
-            GRPC_MAKEFLAGS=--jobs=4 --output-sync=target
-            GRPC_VERSION=v1.63.0
             MAKEFLAGS=${{ inputs.makeflags }}
           context: .
           file: ./Dockerfile
@@ -228,6 +236,14 @@ jobs:
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}
 
+      - 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

.github/workflows/localaibot_automerge.yml

@@ -1,35 +0,0 @@
-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

@@ -1,11 +1,9 @@
 name: Build and Release
 
-on:
-- push
-- pull_request
+on: push
 
 env:
-  GRPC_VERSION: v1.63.0
+  GRPC_VERSION: v1.58.0
 
 permissions:
   contents: write
@@ -16,31 +14,48 @@ concurrency:
 
 jobs:
   build-linux:
+    strategy:
+      matrix:
+        include:
+          - build: 'avx2'
+            defines: ''
+          - build: 'avx'
+            defines: '-DLLAMA_AVX2=OFF'
+          - build: 'avx512'
+            defines: '-DLLAMA_AVX512=ON'
+          - build: 'cuda12'
+            defines: ''
+          - build: 'cuda11'
+            defines: ''
     runs-on: ubuntu-latest
     steps:
       - name: Clone
         uses: actions/checkout@v4
         with:
           submodules: true
-      - uses: actions/setup-go@v5
+      - uses: actions/setup-go@v4
         with:
           go-version: '1.21.x'
           cache: false
       - name: Dependencies
         run: |
           sudo apt-get update
-          sudo apt-get install build-essential ffmpeg protobuf-compiler
+          sudo apt-get install build-essential ffmpeg
       - name: Install CUDA Dependencies
+        if: ${{ matrix.build == 'cuda12' || matrix.build == 'cuda11' }}
         run: |
+          if [ "${{ matrix.build }}" == "cuda12" ]; then
+            export CUDA_VERSION=12-3
+          else
+            export CUDA_VERSION=11-7
+          fi
           curl -O https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb
           sudo dpkg -i cuda-keyring_1.1-1_all.deb
           sudo apt-get update
           sudo apt-get install -y cuda-nvcc-${CUDA_VERSION} libcublas-dev-${CUDA_VERSION}
-        env:
-          CUDA_VERSION: 12-3
       - name: Cache grpc
         id: cache-grpc
-        uses: actions/cache@v4
+        uses: actions/cache@v3
         with:
           path: grpc
           key: ${{ runner.os }}-grpc-${{ env.GRPC_VERSION }}
@@ -56,18 +71,23 @@ jobs:
           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
-          export PATH=/usr/local/cuda/bin:$PATH
-          make dist
-      - uses: actions/upload-artifact@v4
+          if [ "${{ matrix.build }}" == "cuda12" ] || [ "${{ matrix.build }}" == "cuda11" ]; then
+            export BUILD_TYPE=cublas
+            export PATH=/usr/local/cuda/bin:$PATH
+            make dist
+          else
+            STATIC=true make dist
+          fi
+      - uses: actions/upload-artifact@v3
         with:
-          name: LocalAI-linux
+          name: ${{ matrix.build }}
           path: release/
       - name: Release
-        uses: softprops/action-gh-release@v2
+        uses: softprops/action-gh-release@v1
         if: startsWith(github.ref, 'refs/tags/')
         with:
           files: |
@@ -80,54 +100,66 @@ jobs:
         uses: actions/checkout@v4
         with:
           submodules: true
-      - uses: actions/setup-go@v5
+      - uses: actions/setup-go@v4
         with:
           go-version: '1.21.x'
           cache: false
       - name: Dependencies
         run: |
-          sudo apt-get install -y --no-install-recommends libopencv-dev protobuf-compiler
-          go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@latest
-          go install google.golang.org/protobuf/cmd/protoc-gen-go@latest
+          sudo apt-get install -y --no-install-recommends libopencv-dev
       - name: Build stablediffusion
         run: |
-          export PATH=$PATH:$GOPATH/bin
           make backend-assets/grpc/stablediffusion
           mkdir -p release && cp backend-assets/grpc/stablediffusion release
-      - uses: actions/upload-artifact@v4
+      - uses: actions/upload-artifact@v3
         with:
           name: stablediffusion
           path: release/
+      - name: Release
+        uses: softprops/action-gh-release@v1
+        if: startsWith(github.ref, 'refs/tags/')
+        with:
+          files: |
+            release/*
 
-  build-macOS-arm64:
-    runs-on: macos-14
+  build-macOS:
+    strategy:
+      matrix:
+        include:
+          - build: 'avx2'
+            defines: ''
+          - build: 'avx'
+            defines: '-DLLAMA_AVX2=OFF'
+          - build: 'avx512'
+            defines: '-DLLAMA_AVX512=ON'
+    runs-on: macOS-latest
     steps:
       - name: Clone
         uses: actions/checkout@v4
         with:
           submodules: true
-      - uses: actions/setup-go@v5
+      - uses: actions/setup-go@v4
         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
+      - uses: actions/upload-artifact@v3
         with:
-          name: LocalAI-MacOS-arm64
+          name: ${{ matrix.build }}
           path: release/
       - name: Release
-        uses: softprops/action-gh-release@v2
+        uses: softprops/action-gh-release@v1
         if: startsWith(github.ref, 'refs/tags/')
         with:
           files: |

.github/workflows/secscan.yaml

@@ -14,17 +14,14 @@ jobs:
       GO111MODULE: on
     steps:
       - name: Checkout Source
-        uses: actions/checkout@v4
-        if: ${{ github.actor != 'dependabot[bot]' }}
+        uses: actions/checkout@v3
       - 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
+        uses: github/codeql-action/upload-sarif@v2
         with:
           # Path to SARIF file relative to the root of the repository
-          sarif_file: results.sarif
+          sarif_file: results.sarif

.github/workflows/test-extra.yml

@@ -25,14 +25,21 @@ jobs:
         run: |
           sudo apt-get update
           sudo apt-get install build-essential ffmpeg
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          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
-          pip install --user grpcio-tools==1.63.0
           
+          sudo rm -rfv /usr/bin/conda || true
+
       - name: Test transformers
         run: |
+           export PATH=$PATH:/opt/conda/bin
            make --jobs=5 --output-sync=target -C backend/python/transformers
            make --jobs=5 --output-sync=target -C backend/python/transformers test
 
@@ -47,40 +54,24 @@ jobs:
         run: |
           sudo apt-get update
           sudo apt-get install build-essential ffmpeg
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          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
-          pip install --user grpcio-tools==1.63.0
           
+          sudo rm -rfv /usr/bin/conda || true
+
       - name: Test sentencetransformers
         run: |
+           export PATH=$PATH:/opt/conda/bin
            make --jobs=5 --output-sync=target -C backend/python/sentencetransformers
            make --jobs=5 --output-sync=target -C backend/python/sentencetransformers test
 
-
-  tests-rerankers:
-    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
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
-          sudo apt-get install -y libopencv-dev
-          pip install --user grpcio-tools==1.63.0
-
-      - name: Test rerankers
-        run: |
-           make --jobs=5 --output-sync=target -C backend/python/rerankers
-           make --jobs=5 --output-sync=target -C backend/python/rerankers test
-
   tests-diffusers:
     runs-on: ubuntu-latest
     steps:
@@ -91,38 +82,25 @@ jobs:
       - name: Dependencies
         run: |
           sudo apt-get update
-          sudo apt-get install -y build-essential ffmpeg
-          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          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
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          pip install --user grpcio-tools==1.63.0
+          
+          sudo rm -rfv /usr/bin/conda || true
+
       - name: Test diffusers
         run: |
-          make --jobs=5 --output-sync=target -C backend/python/diffusers
-          make --jobs=5 --output-sync=target -C backend/python/diffusers test
+           export PATH=$PATH:/opt/conda/bin
+           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
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
-          sudo apt-get install -y libopencv-dev
-          pip install --user grpcio-tools==1.63.0
-
-      - name: Test parler-tts
-        run: |
-           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
@@ -135,14 +113,21 @@ jobs:
         run: |
           sudo apt-get update
           sudo apt-get install build-essential ffmpeg
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          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
-          pip install --user grpcio-tools==1.63.0
+          
+          sudo rm -rfv /usr/bin/conda || true
 
       - name: Test transformers-musicgen
         run: |
+           export PATH=$PATH:/opt/conda/bin
            make --jobs=5 --output-sync=target -C backend/python/transformers-musicgen
            make --jobs=5 --output-sync=target -C backend/python/transformers-musicgen test
 
@@ -159,14 +144,21 @@ jobs:
   #       run: |
   #         sudo apt-get update
   #         sudo apt-get install build-essential ffmpeg
-  #         # Install UV
-  #         curl -LsSf https://astral.sh/uv/install.sh | sh
-  #         sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+  #         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
-  #         pip install --user grpcio-tools==1.63.0
+          
+  #         sudo rm -rfv /usr/bin/conda || true
 
   #     - 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
 
@@ -223,14 +215,21 @@ jobs:
   #       run: |
   #         sudo apt-get update
   #         sudo apt-get install build-essential ffmpeg
-  #         # Install UV
-  #         curl -LsSf https://astral.sh/uv/install.sh | sh
-  #         sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+  #         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
-  #         pip install --user grpcio-tools==1.63.0
+          
+  #         sudo rm -rfv /usr/bin/conda || true
 
   #     - name: Test bark
   #       run: |
+  #          export PATH=$PATH:/opt/conda/bin
   #          make --jobs=5 --output-sync=target -C backend/python/bark
   #          make --jobs=5 --output-sync=target -C backend/python/bark test
 
@@ -248,13 +247,19 @@ jobs:
   #       run: |
   #         sudo apt-get update
   #         sudo apt-get install build-essential ffmpeg
-  #         # Install UV
-  #         curl -LsSf https://astral.sh/uv/install.sh | sh
-  #         sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+  #         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
-  #         pip install --user grpcio-tools==1.63.0
+  #         sudo rm -rfv /usr/bin/conda || true
   #     - name: Test vllm
   #       run: |
+  #          export PATH=$PATH:/opt/conda/bin
   #          make --jobs=5 --output-sync=target -C backend/python/vllm
   #          make --jobs=5 --output-sync=target -C backend/python/vllm test
   tests-vallex:
@@ -268,13 +273,19 @@ jobs:
         run: |
           sudo apt-get update
           sudo apt-get install build-essential ffmpeg
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
-          sudo apt-get install -y libopencv-dev
-          pip install --user grpcio-tools==1.63.0
+          curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
+             sudo install -o root -g root -m 644 conda.gpg /usr/share/keyrings/conda-archive-keyring.gpg && \
+              gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \
+             sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" > /etc/apt/sources.list.d/conda.list' && \
+             sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
+             sudo apt-get update && \
+             sudo apt-get install -y conda
+          sudo apt-get install -y ca-certificates cmake curl patch
+          sudo apt-get install -y libopencv-dev    
+          sudo rm -rfv /usr/bin/conda || true
       - name: Test vall-e-x
         run: |
+           export PATH=$PATH:/opt/conda/bin
            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
 
@@ -289,11 +300,18 @@ jobs:
         run: |
           sudo apt-get update
           sudo apt-get install build-essential ffmpeg
-          sudo apt-get install -y ca-certificates cmake curl patch espeak espeak-ng python3-pip
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          pip install --user grpcio-tools==1.63.0
+          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 espeak espeak-ng          
+          sudo rm -rfv /usr/bin/conda || true
+
       - name: Test coqui
         run: |
-          make --jobs=5 --output-sync=target -C backend/python/coqui
-          make --jobs=5 --output-sync=target -C backend/python/coqui test
+           export PATH=$PATH:/opt/conda/bin
+           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

@@ -10,7 +10,7 @@ on:
       - '*'
 
 env:
-  GRPC_VERSION: v1.63.0
+  GRPC_VERSION: v1.58.0
 
 concurrency:
   group: ci-tests-${{ github.head_ref || github.ref }}-${{ github.repository }}
@@ -57,10 +57,10 @@ jobs:
           df -h
       - name: Clone
         uses: actions/checkout@v4
-        with:
+        with: 
           submodules: true
       - name: Setup Go ${{ matrix.go-version }}
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v4
         with:
           go-version: ${{ matrix.go-version }}
           cache: false
@@ -70,35 +70,17 @@ jobs:
       - name: Dependencies
         run: |
           sudo apt-get update
-          sudo apt-get install build-essential curl ffmpeg
+          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 && \
+              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
-          # Install UV
-          curl -LsSf https://astral.sh/uv/install.sh | sh
-          sudo apt-get install -y ca-certificates cmake patch python3-pip unzip
+          sudo apt-get install -y ca-certificates cmake curl patch
           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
-
-          curl -O https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb
-          sudo dpkg -i cuda-keyring_1.1-1_all.deb
-          sudo apt-get update
-          sudo apt-get install -y cuda-nvcc-${CUDA_VERSION} libcublas-dev-${CUDA_VERSION}
-          export CUDACXX=/usr/local/cuda/bin/nvcc
-
-          go install google.golang.org/protobuf/cmd/protoc-gen-go@latest
-          go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@latest
-
-          # The python3-grpc-tools package in 22.04 is too old
-          pip install --user grpcio-tools
-
+          
           sudo rm -rfv /usr/bin/conda || true
           PATH=$PATH:/opt/conda/bin make -C backend/python/sentencetransformers
 
@@ -107,12 +89,10 @@ 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)
-          PATH="$PATH:/root/go/bin" GO_TAGS="stablediffusion tts" GRPC_BACKENDS=backend-assets/grpc/stablediffusion make build
-        env:
-          CUDA_VERSION: 12-3
+          GO_TAGS="stablediffusion tts" GRPC_BACKENDS=backend-assets/grpc/stablediffusion make build
       - name: Cache grpc
         id: cache-grpc
-        uses: actions/cache@v4
+        uses: actions/cache@v3
         with:
           path: grpc
           key: ${{ runner.os }}-grpc-${{ env.GRPC_VERSION }}
@@ -128,14 +108,11 @@ jobs:
           cd grpc && cd cmake/build && sudo make --jobs 5 install
       - name: Test
         run: |
-          PATH="$PATH:/root/go/bin" GO_TAGS="stablediffusion tts" make --jobs 5 --output-sync=target test
+          GO_TAGS="stablediffusion tts" make --jobs 5 --output-sync=target test
       - name: Setup tmate session if tests fail
         if: ${{ failure() }}
-        uses: mxschmitt/action-tmate@v3.18
-        with:
-          detached: true
-          connect-timeout-seconds: 180
-          limit-access-to-actor: true
+        uses: mxschmitt/action-tmate@v3
+        timeout-minutes: 5
 
   tests-aio-container:
     runs-on: ubuntu-latest
@@ -174,7 +151,7 @@ jobs:
           df -h
       - name: Clone
         uses: actions/checkout@v4
-        with:
+        with: 
           submodules: true
       - name: Build images
         run: |
@@ -186,11 +163,8 @@ jobs:
             make run-e2e-aio
       - name: Setup tmate session if tests fail
         if: ${{ failure() }}
-        uses: mxschmitt/action-tmate@v3.18
-        with:
-          detached: true
-          connect-timeout-seconds: 180
-          limit-access-to-actor: true
+        uses: mxschmitt/action-tmate@v3
+        timeout-minutes: 5
 
   tests-apple:
     runs-on: macOS-14
@@ -200,10 +174,10 @@ jobs:
     steps:
       - name: Clone
         uses: actions/checkout@v4
-        with:
+        with: 
           submodules: true
       - name: Setup Go ${{ matrix.go-version }}
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v4
         with:
           go-version: ${{ matrix.go-version }}
           cache: false
@@ -212,8 +186,7 @@ jobs:
         run: go version
       - name: Dependencies
         run: |
-          brew install protobuf grpc make protoc-gen-go protoc-gen-go-grpc
-          pip install --user grpcio-tools==1.63.0
+          brew install protobuf grpc make
       - name: Test
         run: |
           export C_INCLUDE_PATH=/usr/local/include
@@ -223,8 +196,5 @@ jobs:
           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@v3.18
-        with:
-          detached: true
-          connect-timeout-seconds: 180
-          limit-access-to-actor: true
+        uses: mxschmitt/action-tmate@v3
+        timeout-minutes: 5

.github/workflows/update_swagger.yaml

@@ -1,31 +0,0 @@
-name: Update swagger
-on:
-  schedule:
-    - cron: 0 20 * * *
-  workflow_dispatch:
-jobs:
-  swagger:
-    strategy:
-      fail-fast: false
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v4
-      - uses: actions/setup-go@v5
-        with:
-          go-version: 'stable'
-      - run: |
-          go install github.com/swaggo/swag/cmd/swag@latest
-      - name: Bump swagger 🔧
-        run: |
-          make swagger
-      - name: Create Pull Request
-        uses: peter-evans/create-pull-request@v6
-        with:
-          token: ${{ secrets.UPDATE_BOT_TOKEN }}
-          push-to-fork: ci-forks/LocalAI
-          commit-message: 'feat(swagger): update swagger'
-          title: 'feat(swagger): update swagger'
-          branch: "update/swagger"
-          body:  Update swagger
-          signoff: true
-

.github/workflows/yaml-check.yml

@@ -1,18 +0,0 @@
-name: 'Yamllint GitHub Actions'
-on:
-  - pull_request
-jobs:
-  yamllint:
-    name: 'Yamllint'
-    runs-on: ubuntu-latest
-    steps:
-      - name: 'Checkout'
-        uses: actions/checkout@master
-      - name: 'Yamllint'
-        uses: karancode/yamllint-github-action@master
-        with:
-          yamllint_file_or_dir: 'gallery'
-          yamllint_strict: false
-          yamllint_comment: true
-        env:
-          GITHUB_ACCESS_TOKEN: ${{ secrets.GITHUB_TOKEN }}

.gitignore

@@ -39,14 +39,3 @@ backend-assets/*
 !backend-assets/.keep
 prepare
 /ggml-metal.metal
-
-# Protobuf generated files
-*.pb.go
-*pb2.py
-*pb2_grpc.py
-
-# SonarQube
-.scannerwork
-
-# backend virtual environments
-**/venv

.yamllint

@@ -1,4 +0,0 @@
-extends: default
-
-rules:
-    line-length: disable

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.
 
@@ -29,9 +29,8 @@ 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 ( see https://localai.io/basics/build/#build-localai-locally )
-4. Build LocalAI: `make build`
-5. Run LocalAI: `./local-ai`
+3. Install the required dependencies: `make prepare`
+4. Run LocalAI: `make run`
 
 ## Contributing
 
@@ -60,29 +59,14 @@ 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 or create a new issue. The documentation is available under `docs/` https://github.com/mudler/LocalAI/tree/master/docs
- 
+- 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)
+
 ## 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

@@ -1,45 +1,30 @@
 ARG IMAGE_TYPE=extras
 ARG BASE_IMAGE=ubuntu:22.04
-ARG GRPC_BASE_IMAGE=${BASE_IMAGE}
 
-# The requirements-core target is common to all images.  It should not be placed in requirements-core unless every single build will use it.
-FROM ${BASE_IMAGE} AS requirements-core
+# extras or core
+FROM ${BASE_IMAGE} as requirements-core
 
 USER root
 
 ARG GO_VERSION=1.21.7
+ARG BUILD_TYPE
+ARG CUDA_MAJOR_VERSION=11
+ARG CUDA_MINOR_VERSION=7
 ARG TARGETARCH
 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,rerankers:/build/backend/python/rerankers/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"
+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"
 
 ARG GO_TAGS="stablediffusion tinydream tts"
 
 RUN apt-get update && \
-    apt-get install -y --no-install-recommends \
-        build-essential \
-        ca-certificates \
-        cmake \
-        curl \
-        git \
-        python3-pip \
-        python-is-python3 \
-        unzip && \
-    apt-get clean && \
-    rm -rf /var/lib/apt/lists/* && \
-    pip install --upgrade pip
+    apt-get install -y ca-certificates curl patch pip cmake git && 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 $PATH:/root/go/bin:/usr/local/go/bin
-
-# Install grpc compilers
-RUN go install google.golang.org/protobuf/cmd/protoc-gen-go@latest && \
-    go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@latest
-
-# Install grpcio-tools (the version in 22.04 is too old)
-RUN pip install --user grpcio-tools
+RUN curl -L -s https://go.dev/dl/go$GO_VERSION.linux-$TARGETARCH.tar.gz | tar -C /usr/local -xz
+ENV PATH $PATH:/usr/local/go/bin
 
 COPY --chmod=644 custom-ca-certs/* /usr/local/share/ca-certificates/
 RUN update-ca-certificates
@@ -48,6 +33,16 @@ RUN update-ca-certificates
 RUN echo "Target Architecture: $TARGETARCH"
 RUN echo "Target Variant: $TARGETVARIANT"
 
+# CuBLAS requirements
+RUN if [ "${BUILD_TYPE}" = "cublas" ]; then \
+    apt-get install -y software-properties-common && \
+    curl -O https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb && \
+    dpkg -i cuda-keyring_1.1-1_all.deb && \
+    rm -f cuda-keyring_1.1-1_all.deb && \
+    apt-get update && \
+    apt-get install -y cuda-nvcc-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION} libcurand-dev-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION} libcublas-dev-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION} libcusparse-dev-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION} libcusolver-dev-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION}  && apt-get clean \
+    ; fi
+
 # Cuda
 ENV PATH /usr/local/cuda/bin:${PATH}
 
@@ -55,12 +50,10 @@ ENV PATH /usr/local/cuda/bin:${PATH}
 ENV PATH /opt/rocm/bin:${PATH}
 
 # OpenBLAS requirements and stable diffusion
-RUN apt-get update && \
-    apt-get install -y --no-install-recommends \
-        libopenblas-dev \
-        libopencv-dev && \
-    apt-get clean && \
-    rm -rf /var/lib/apt/lists/*
+RUN apt-get install -y \
+    libopenblas-dev \
+    libopencv-dev \ 
+    && apt-get clean
 
 # Set up OpenCV
 RUN ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
@@ -73,116 +66,56 @@ RUN test -n "$TARGETARCH" \
 ###################################
 ###################################
 
-# The requirements-extras target is for any builds with IMAGE_TYPE=extras. It should not be placed in this target unless every IMAGE_TYPE=extras build will use it
-FROM requirements-core AS requirements-extras
+FROM requirements-core as requirements-extras
+
+RUN 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 && \
+    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 && \
+    apt-get update && \
+    apt-get install -y conda && apt-get clean
 
-RUN curl -LsSf https://astral.sh/uv/install.sh | sh
 ENV PATH="/root/.cargo/bin:${PATH}"
+RUN apt-get install -y python3-pip && apt-get clean
+RUN pip install --upgrade pip
 
 RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
-RUN apt-get update && \
-    apt-get install -y --no-install-recommends \
-        espeak-ng \
-        espeak \
-        python3-dev \
-        python3-venv && \
-    apt-get clean && \
-    rm -rf /var/lib/apt/lists/*
+RUN apt-get install -y espeak-ng espeak && apt-get clean
 
-###################################
-###################################
-
-# The requirements-drivers target is for BUILD_TYPE specific items.  If you need to install something specific to CUDA, or specific to ROCM, it goes here.
-# This target will be built on top of requirements-core or requirements-extras as retermined by the IMAGE_TYPE build-arg
-FROM requirements-${IMAGE_TYPE} AS requirements-drivers
-
-ARG BUILD_TYPE
-ARG CUDA_MAJOR_VERSION=11
-ARG CUDA_MINOR_VERSION=7
-
-ENV BUILD_TYPE=${BUILD_TYPE}
-
-# CuBLAS requirements
-RUN if [ "${BUILD_TYPE}" = "cublas" ]; then \
-        apt-get update && \
-        apt-get install -y  --no-install-recommends \
-            software-properties-common && \
-        curl -O https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb && \
-        dpkg -i cuda-keyring_1.1-1_all.deb && \
-        rm -f cuda-keyring_1.1-1_all.deb && \
-        apt-get update && \
-        apt-get install -y --no-install-recommends \
-            cuda-nvcc-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION} \
-            libcurand-dev-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION} \
-            libcublas-dev-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION} \
-            libcusparse-dev-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION} \
-            libcusolver-dev-${CUDA_MAJOR_VERSION}-${CUDA_MINOR_VERSION} && \
-        apt-get clean && \
-        rm -rf /var/lib/apt/lists/* \
-    ; fi
-
-# If we are building with clblas support, we need the libraries for the builds
-RUN if [ "${BUILD_TYPE}" = "clblas" ]; then \
-        apt-get update && \
-        apt-get install -y --no-install-recommends \
-            libclblast-dev && \
-        apt-get clean && \
-        rm -rf /var/lib/apt/lists/* \
-    ; fi
-
-RUN if [ "${BUILD_TYPE}" = "hipblas" ]; then \
-        apt-get update && \
-        apt-get install -y --no-install-recommends \
-            hipblas-dev \
-            rocblas-dev && \
-        apt-get clean && \
-        rm -rf /var/lib/apt/lists/* && \
-        # I have no idea why, but the ROCM lib packages don't trigger ldconfig after they install, which results in local-ai and others not being able
-        # to locate the libraries. We run ldconfig ourselves to work around this packaging deficiency
-        ldconfig \
+RUN if [ ! -e /usr/bin/python ]; then \
+	  ln -s /usr/bin/python3 /usr/bin/python \
     ; fi
 
 ###################################
 ###################################
 
-# The grpc target does one thing, it builds and installs GRPC.  This is in it's own layer so that it can be effectively cached by CI.
-# You probably don't need to change anything here, and if you do, make sure that CI is adjusted so that the cache continues to work.
-FROM ${GRPC_BASE_IMAGE} AS grpc
+FROM ${BASE_IMAGE} as grpc
 
-# This is a bit of a hack, but it's required in order to be able to effectively cache this layer in CI
-ARG GRPC_MAKEFLAGS="-j4 -Otarget"
+ARG MAKEFLAGS
 ARG GRPC_VERSION=v1.58.0
 
-ENV MAKEFLAGS=${GRPC_MAKEFLAGS}
+ENV MAKEFLAGS=${MAKEFLAGS}
 
 WORKDIR /build
 
 RUN apt-get update && \
-    apt-get install -y --no-install-recommends \
-        ca-certificates \
-        build-essential \
-        cmake \
-        git && \
+    apt-get install -y g++ cmake git && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/*
 
-# We install GRPC to a different prefix here so that we can copy in only the build artifacts later
-# saves several hundred MB on the final docker image size vs copying in the entire GRPC source tree
-# and running make install in the target container
-RUN git clone --recurse-submodules --jobs 4 -b ${GRPC_VERSION} --depth 1 --shallow-submodules https://github.com/grpc/grpc && \
-    mkdir -p /build/grpc/cmake/build && \
-    cd /build/grpc/cmake/build && \
-    cmake -DgRPC_INSTALL=ON -DgRPC_BUILD_TESTS=OFF -DCMAKE_INSTALL_PREFIX:PATH=/opt/grpc ../.. && \
-    make && \
-    make install && \
-    rm -rf /build
+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
 
 ###################################
 ###################################
 
-# The builder target compiles LocalAI. This target is not the target that will be uploaded to the registry.
-# Adjustments to the build process should likely be made here.
-FROM requirements-drivers AS builder
+FROM requirements-${IMAGE_TYPE} as builder
 
 ARG GO_TAGS="stablediffusion tts"
 ARG GRPC_BACKENDS
@@ -200,43 +133,39 @@ WORKDIR /build
 COPY . .
 COPY .git .
 RUN echo "GO_TAGS: $GO_TAGS"
-
 RUN make prepare
 
-# We need protoc installed, and the version in 22.04 is too old.  We will create one as part installing the GRPC build below
-# but that will also being in a newer version of absl which stablediffusion cannot compile with.  This version of protoc is only
-# here so that we can generate the grpc code for the stablediffusion build
-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
+# If we are building with clblas support, we need the libraries for the builds
+RUN if [ "${BUILD_TYPE}" = "clblas" ]; then \
+    apt-get update && \
+    apt-get install -y libclblast-dev && \
+    apt-get clean \
+    ; fi
 
 # stablediffusion does not tolerate a newer version of abseil, build it first
 RUN GRPC_BACKENDS=backend-assets/grpc/stablediffusion make build
 
-# Install the pre-built GRPC
-COPY --from=grpc /opt/grpc /usr/local
+COPY --from=grpc /build/grpc ./grpc/
+
+RUN cd /build/grpc/cmake/build && make install
 
 # Rebuild with defaults backends
-WORKDIR /build
 RUN make build
 
 RUN if [ ! -d "/build/sources/go-piper/piper-phonemize/pi/lib/" ]; then \
-        mkdir -p /build/sources/go-piper/piper-phonemize/pi/lib/ \
-        touch /build/sources/go-piper/piper-phonemize/pi/lib/keep \
+    mkdir -p /build/sources/go-piper/piper-phonemize/pi/lib/ \
+    touch /build/sources/go-piper/piper-phonemize/pi/lib/keep \
     ; fi
 
 ###################################
 ###################################
 
-# This is the final target. The result of this target will be the image uploaded to the registry.
-# If you cannot find a more suitable place for an addition, this layer is a suitable place for it.
-FROM requirements-drivers
+FROM requirements-${IMAGE_TYPE}
 
 ARG FFMPEG
 ARG BUILD_TYPE
 ARG TARGETARCH
 ARG IMAGE_TYPE=extras
-ARG EXTRA_BACKENDS
 ARG MAKEFLAGS
 
 ENV BUILD_TYPE=${BUILD_TYPE}
@@ -248,14 +177,18 @@ ARG CUDA_MAJOR_VERSION=11
 ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility
 ENV NVIDIA_REQUIRE_CUDA="cuda>=${CUDA_MAJOR_VERSION}.0"
 ENV NVIDIA_VISIBLE_DEVICES=all
+ENV PIP_CACHE_PURGE=true
 
 # Add FFmpeg
 RUN if [ "${FFMPEG}" = "true" ]; then \
-        apt-get update && \
-        apt-get install -y --no-install-recommends \
-            ffmpeg && \
-        apt-get clean && \
-        rm -rf /var/lib/apt/lists/* \
+    apt-get install -y ffmpeg && apt-get clean \
+    ; fi
+
+# Add OpenCL
+RUN if [ "${BUILD_TYPE}" = "clblas" ]; then \
+    apt-get update && \
+    apt-get install -y libclblast1 && \
+    apt-get clean \
     ; fi
 
 WORKDIR /build
@@ -267,9 +200,9 @@ WORKDIR /build
 COPY . .
 
 COPY --from=builder /build/sources ./sources/
-COPY --from=grpc /opt/grpc /usr/local
+COPY --from=grpc /build/grpc ./grpc/
 
-RUN make prepare-sources
+RUN make prepare-sources && cd /build/grpc/cmake/build && make install && rm -rf grpc
 
 # Copy the binary
 COPY --from=builder /build/local-ai ./
@@ -280,58 +213,45 @@ COPY --from=builder /build/sources/go-piper/piper-phonemize/pi/lib/* /usr/lib/
 # do not let stablediffusion rebuild (requires an older version of absl)
 COPY --from=builder /build/backend-assets/grpc/stablediffusion ./backend-assets/grpc/stablediffusion
 
-# Change the shell to bash so we can use [[ tests below
-SHELL ["/bin/bash", "-c"]
-# We try to strike a balance between individual layer size (as that affects total push time) and total image size
-# Splitting the backends into more groups with fewer items results in a larger image, but a smaller size for the largest layer
-# Splitting the backends into fewer groups with more items results in a smaller image, but a larger size for the largest layer
-
-RUN if [[ ( "${EXTRA_BACKENDS}" =~ "coqui" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/coqui \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "parler-tts" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/parler-tts \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "diffusers" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/diffusers \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "transformers-musicgen" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/transformers-musicgen \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "exllama1" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/exllama \
+## Duplicated from Makefile to avoid having a big layer that's hard to push
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/autogptq \
     ; fi
-
-RUN if [[ ( "${EXTRA_BACKENDS}" =~ "vall-e-x" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/vall-e-x \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "petals" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/petals \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "sentencetransformers" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/sentencetransformers \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "exllama2" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/exllama2 \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "transformers" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/transformers \
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/bark \
     ; fi
-
-RUN if [[ ( "${EXTRA_BACKENDS}" =~ "vllm" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/vllm \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "autogptq" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/autogptq \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "bark" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/bark \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "rerankers" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/rerankers \
-    ; fi && \
-    if [[ ( "${EXTRA_BACKENDS}" =~ "mamba" || -z "${EXTRA_BACKENDS}" ) && "$IMAGE_TYPE" == "extras" ]]; then \
-        make -C backend/python/mamba \
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/diffusers \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/vllm \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/mamba \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/sentencetransformers \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/transformers \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/vall-e-x \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/exllama \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/exllama2 \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/petals \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/transformers-musicgen \
+    ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+    make -C backend/python/coqui \
     ; fi
 
 # Make sure the models directory exists
@@ -339,7 +259,7 @@ RUN mkdir -p /build/models
 
 # Define the health check command
 HEALTHCHECK --interval=1m --timeout=10m --retries=10 \
-  CMD curl -f ${HEALTHCHECK_ENDPOINT} || exit 1
+  CMD curl -f $HEALTHCHECK_ENDPOINT || exit 1
   
 VOLUME /build/models
 EXPOSE 8080

Makefile

@@ -5,7 +5,7 @@ BINARY_NAME=local-ai
 
 # llama.cpp versions
 GOLLAMA_STABLE_VERSION?=2b57a8ae43e4699d3dc5d1496a1ccd42922993be
-CPPLLAMA_VERSION?=dc685be46622a8fabfd57cfa804237c8f15679b8
+CPPLLAMA_VERSION?=1b67731e184e27a465b8c5476061294a4af668ea
 
 # 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?=4ef8d9f44eb402c528ab6d990ab50a9f4f666347
+WHISPER_CPP_VERSION?=8f253ef3af1c62c04316ba4afa7145fc4d701a8c
 
 # bert.cpp version
 BERT_VERSION?=6abe312cded14042f6b7c3cd8edf082713334a4d
@@ -25,10 +25,10 @@ BERT_VERSION?=6abe312cded14042f6b7c3cd8edf082713334a4d
 PIPER_VERSION?=9d0100873a7dbb0824dfea40e8cec70a1b110759
 
 # stablediffusion version
-STABLEDIFFUSION_VERSION?=4a3cd6aeae6f66ee57eae9a0075f8c58c3a6a38f
+STABLEDIFFUSION_VERSION?=362df9da29f882dbf09ade61972d16a1f53c3485
 
 # tinydream version
-TINYDREAM_VERSION?=c04fa463ace9d9a6464313aa5f9cd0f953b6c057
+TINYDREAM_VERSION?=22a12a4bc0ac5455856f28f3b771331a551a4293
 
 export BUILD_TYPE?=
 export STABLE_BUILD_TYPE?=$(BUILD_TYPE)
@@ -38,7 +38,7 @@ CGO_LDFLAGS?=
 CGO_LDFLAGS_WHISPER?=
 CUDA_LIBPATH?=/usr/local/cuda/lib64/
 GO_TAGS?=
-BUILD_ID?=
+BUILD_ID?=git
 
 TEST_DIR=/tmp/test
 
@@ -70,7 +70,7 @@ UNAME_S := $(shell uname -s)
 endif
 
 ifeq ($(OS),Darwin)
-
+	
 	ifeq ($(OSX_SIGNING_IDENTITY),)
 		OSX_SIGNING_IDENTITY := $(shell security find-identity -v -p codesigning | grep '"' | head -n 1 | sed -E 's/.*"(.*)"/\1/')
 	endif
@@ -99,7 +99,7 @@ endif
 ifeq ($(BUILD_TYPE),cublas)
 	CGO_LDFLAGS+=-lcublas -lcudart -L$(CUDA_LIBPATH)
 	export LLAMA_CUBLAS=1
-	export WHISPER_CUDA=1
+	export WHISPER_CUBLAS=1
 	CGO_LDFLAGS_WHISPER+=-L$(CUDA_LIBPATH)/stubs/ -lcuda
 endif
 
@@ -152,11 +152,9 @@ ifeq ($(findstring tts,$(GO_TAGS)),tts)
 	OPTIONAL_GRPC+=backend-assets/grpc/piper
 endif
 
-ALL_GRPC_BACKENDS=backend-assets/grpc/huggingface
+ALL_GRPC_BACKENDS=backend-assets/grpc/langchain-huggingface
 ALL_GRPC_BACKENDS+=backend-assets/grpc/bert-embeddings
-ALL_GRPC_BACKENDS+=backend-assets/grpc/llama-cpp-avx
-ALL_GRPC_BACKENDS+=backend-assets/grpc/llama-cpp-avx2
-ALL_GRPC_BACKENDS+=backend-assets/grpc/llama-cpp-fallback
+ALL_GRPC_BACKENDS+=backend-assets/grpc/llama-cpp
 ALL_GRPC_BACKENDS+=backend-assets/grpc/llama-ggml
 ALL_GRPC_BACKENDS+=backend-assets/grpc/gpt4all
 ALL_GRPC_BACKENDS+=backend-assets/grpc/rwkv
@@ -181,20 +179,20 @@ endif
 all: help
 
 ## BERT embeddings
-sources/go-bert.cpp:
-	git clone --recurse-submodules https://github.com/go-skynet/go-bert.cpp sources/go-bert.cpp
-	cd sources/go-bert.cpp && git checkout -b build $(BERT_VERSION) && git submodule update --init --recursive --depth 1
+sources/go-bert:
+	git clone --recurse-submodules https://github.com/go-skynet/go-bert.cpp sources/go-bert
+	cd sources/go-bert && git checkout -b build $(BERT_VERSION) && git submodule update --init --recursive --depth 1
 
-sources/go-bert.cpp/libgobert.a: sources/go-bert.cpp
-	$(MAKE) -C sources/go-bert.cpp libgobert.a
+sources/go-bert/libgobert.a: sources/go-bert
+	$(MAKE) -C sources/go-bert libgobert.a
 
-## go-llama.cpp
-sources/go-llama.cpp:
-	git clone --recurse-submodules https://github.com/go-skynet/go-llama.cpp sources/go-llama.cpp
-	cd sources/go-llama.cpp && git checkout -b build $(GOLLAMA_STABLE_VERSION) && git submodule update --init --recursive --depth 1
+## go-llama-ggml
+sources/go-llama-ggml:
+	git clone --recurse-submodules https://github.com/go-skynet/go-llama.cpp sources/go-llama-ggml
+	cd sources/go-llama-ggml && git checkout -b build $(GOLLAMA_STABLE_VERSION) && git submodule update --init --recursive --depth 1
 
-sources/go-llama.cpp/libbinding.a: sources/go-llama.cpp
-	$(MAKE) -C sources/go-llama.cpp BUILD_TYPE=$(STABLE_BUILD_TYPE) libbinding.a
+sources/go-llama-ggml/libbinding.a: sources/go-llama-ggml
+	$(MAKE) -C sources/go-llama-ggml BUILD_TYPE=$(STABLE_BUILD_TYPE) libbinding.a
 
 ## go-piper
 sources/go-piper:
@@ -213,12 +211,12 @@ sources/gpt4all/gpt4all-bindings/golang/libgpt4all.a: sources/gpt4all
 	$(MAKE) -C sources/gpt4all/gpt4all-bindings/golang/ libgpt4all.a
 
 ## RWKV
-sources/go-rwkv.cpp:
-	git clone --recurse-submodules $(RWKV_REPO) sources/go-rwkv.cpp
-	cd sources/go-rwkv.cpp && git checkout -b build $(RWKV_VERSION) && git submodule update --init --recursive --depth 1
+sources/go-rwkv:
+	git clone --recurse-submodules $(RWKV_REPO) sources/go-rwkv
+	cd sources/go-rwkv && git checkout -b build $(RWKV_VERSION) && git submodule update --init --recursive --depth 1
 
-sources/go-rwkv.cpp/librwkv.a: sources/go-rwkv.cpp
-	cd sources/go-rwkv.cpp && cd rwkv.cpp &&	cmake . -DRWKV_BUILD_SHARED_LIBRARY=OFF &&	cmake --build . && 	cp librwkv.a ..
+sources/go-rwkv/librwkv.a: sources/go-rwkv
+	cd sources/go-rwkv && cd rwkv.cpp &&	cmake . -DRWKV_BUILD_SHARED_LIBRARY=OFF &&	cmake --build . && 	cp librwkv.a ..
 
 ## stable diffusion
 sources/go-stable-diffusion:
@@ -238,24 +236,23 @@ sources/go-tiny-dream/libtinydream.a: sources/go-tiny-dream
 
 ## whisper
 sources/whisper.cpp:
-	git clone https://github.com/ggerganov/whisper.cpp sources/whisper.cpp
+	git clone https://github.com/ggerganov/whisper.cpp.git sources/whisper.cpp
 	cd sources/whisper.cpp && git checkout -b build $(WHISPER_CPP_VERSION) && git submodule update --init --recursive --depth 1
 
 sources/whisper.cpp/libwhisper.a: sources/whisper.cpp
-	cd sources/whisper.cpp && $(MAKE) libwhisper.a
+	cd sources/whisper.cpp && make libwhisper.a
 
-get-sources: sources/go-llama.cpp sources/gpt4all sources/go-piper sources/go-rwkv.cpp sources/whisper.cpp sources/go-bert.cpp sources/go-stable-diffusion sources/go-tiny-dream
+get-sources: sources/go-llama-ggml sources/gpt4all sources/go-piper sources/go-rwkv sources/whisper.cpp sources/go-bert sources/go-stable-diffusion sources/go-tiny-dream
 
 replace:
-	$(GOCMD) mod edit -replace github.com/donomii/go-rwkv.cpp=$(CURDIR)/sources/go-rwkv.cpp
+	$(GOCMD) mod edit -replace github.com/donomii/go-rwkv.cpp=$(CURDIR)/sources/go-rwkv
 	$(GOCMD) mod edit -replace github.com/ggerganov/whisper.cpp=$(CURDIR)/sources/whisper.cpp
 	$(GOCMD) mod edit -replace github.com/ggerganov/whisper.cpp/bindings/go=$(CURDIR)/sources/whisper.cpp/bindings/go
-	$(GOCMD) mod edit -replace github.com/go-skynet/go-bert.cpp=$(CURDIR)/sources/go-bert.cpp
+	$(GOCMD) mod edit -replace github.com/go-skynet/go-bert.cpp=$(CURDIR)/sources/go-bert
 	$(GOCMD) mod edit -replace github.com/M0Rf30/go-tiny-dream=$(CURDIR)/sources/go-tiny-dream
 	$(GOCMD) mod edit -replace github.com/mudler/go-piper=$(CURDIR)/sources/go-piper
 	$(GOCMD) mod edit -replace github.com/mudler/go-stable-diffusion=$(CURDIR)/sources/go-stable-diffusion
 	$(GOCMD) mod edit -replace github.com/nomic-ai/gpt4all/gpt4all-bindings/golang=$(CURDIR)/sources/gpt4all/gpt4all-bindings/golang
-	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(CURDIR)/sources/go-llama.cpp
 
 dropreplace:
 	$(GOCMD) mod edit -dropreplace github.com/donomii/go-rwkv.cpp
@@ -274,12 +271,12 @@ prepare-sources: get-sources replace
 ## GENERIC
 rebuild: ## Rebuilds the project
 	$(GOCMD) clean -cache
-	$(MAKE) -C sources/go-llama.cpp clean
+	$(MAKE) -C sources/go-llama-ggml clean
 	$(MAKE) -C sources/gpt4all/gpt4all-bindings/golang/ clean
-	$(MAKE) -C sources/go-rwkv.cpp clean
+	$(MAKE) -C sources/go-rwkv clean
 	$(MAKE) -C sources/whisper.cpp clean
 	$(MAKE) -C sources/go-stable-diffusion clean
-	$(MAKE) -C sources/go-bert.cpp clean
+	$(MAKE) -C sources/go-bert clean
 	$(MAKE) -C sources/go-piper clean
 	$(MAKE) -C sources/go-tiny-dream clean
 	$(MAKE) build
@@ -292,13 +289,10 @@ 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
-	rm -rf backend/cpp/llama-* || true
 	$(MAKE) dropreplace
-	$(MAKE) protogen-clean
-	rmdir pkg/grpc/proto || true
 
 clean-tests:
 	rm -rf test-models
@@ -314,26 +308,14 @@ build: prepare backend-assets grpcs ## Build the project
 	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_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:
-	STATIC=true $(MAKE) backend-assets/grpc/llama-cpp-avx2
-ifeq ($(OS),Darwin)
-	$(info ${GREEN}I Skip CUDA build on MacOS${RESET})
-else
-	$(MAKE) backend-assets/grpc/llama-cpp-cuda
-endif
-	$(MAKE) build
+dist: build
 	mkdir -p release
-# if BUILD_ID is empty, then we don't append it to the binary name
-ifeq ($(BUILD_ID),)
-	cp $(BINARY_NAME) release/$(BINARY_NAME)-$(OS)-$(ARCH)
-else
 	cp $(BINARY_NAME) release/$(BINARY_NAME)-$(BUILD_ID)-$(OS)-$(ARCH)
-endif
 
 osx-signed: build
 	codesign --deep --force --sign "$(OSX_SIGNING_IDENTITY)" --entitlements "./Entitlements.plist" "./$(BINARY_NAME)"
@@ -434,152 +416,30 @@ 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
 
-.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 rerankers-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 rerankers-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: rerankers-protogen
-rerankers-protogen:
-	$(MAKE) -C backend/python/rerankers protogen
-
-.PHONY: rerankers-protogen-clean
-rerankers-protogen-clean:
-	$(MAKE) -C backend/python/rerankers 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
+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
 
 ## GRPC
 # Note: it is duplicated in the Dockerfile
-prepare-extra-conda-environments: protogen-python
+prepare-extra-conda-environments:
 	$(MAKE) -C backend/python/autogptq
 	$(MAKE) -C backend/python/bark
 	$(MAKE) -C backend/python/coqui
@@ -587,16 +447,14 @@ prepare-extra-conda-environments: protogen-python
 	$(MAKE) -C backend/python/vllm
 	$(MAKE) -C backend/python/mamba
 	$(MAKE) -C backend/python/sentencetransformers
-	$(MAKE) -C backend/python/rerankers
 	$(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: protogen-python
+prepare-test-extra:
 	$(MAKE) -C backend/python/transformers
 	$(MAKE) -C backend/python/diffusers
 
@@ -620,19 +478,19 @@ 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: protogen-go replace
+backend-assets/grpc: replace
 	mkdir -p backend-assets/grpc
 
-backend-assets/grpc/bert-embeddings: sources/go-bert.cpp sources/go-bert.cpp/libgobert.a backend-assets/grpc
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/go-bert.cpp LIBRARY_PATH=$(CURDIR)/sources/go-bert.cpp \
+backend-assets/grpc/bert-embeddings: sources/go-bert sources/go-bert/libgobert.a backend-assets/grpc
+	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/go-bert LIBRARY_PATH=$(CURDIR)/sources/go-bert \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/bert-embeddings ./backend/go/llm/bert/
 
 backend-assets/grpc/gpt4all: sources/gpt4all sources/gpt4all/gpt4all-bindings/golang/libgpt4all.a backend-assets/gpt4all backend-assets/grpc
 	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/gpt4all/gpt4all-bindings/golang/ LIBRARY_PATH=$(CURDIR)/sources/gpt4all/gpt4all-bindings/golang/ \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/gpt4all ./backend/go/llm/gpt4all/
 
-backend-assets/grpc/huggingface: backend-assets/grpc
-	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/huggingface ./backend/go/llm/langchain/
+backend-assets/grpc/langchain-huggingface: backend-assets/grpc
+	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/langchain-huggingface ./backend/go/llm/langchain/
 
 backend/cpp/llama/llama.cpp:
 	LLAMA_VERSION=$(CPPLLAMA_VERSION) $(MAKE) -C backend/cpp/llama llama.cpp
@@ -644,7 +502,7 @@ ADDED_CMAKE_ARGS=-Dabsl_DIR=${INSTALLED_LIB_CMAKE}/absl \
 				 -Dutf8_range_DIR=${INSTALLED_LIB_CMAKE}/utf8_range \
 				 -DgRPC_DIR=${INSTALLED_LIB_CMAKE}/grpc \
 				 -DCMAKE_CXX_STANDARD_INCLUDE_DIRECTORIES=${INSTALLED_PACKAGES}/include
-build-llama-cpp-grpc-server:
+backend/cpp/llama/grpc-server:
 # Conditionally build grpc for the llama backend to use if needed
 ifdef BUILD_GRPC_FOR_BACKEND_LLAMA
 	$(MAKE) -C backend/cpp/grpc build
@@ -653,54 +511,30 @@ ifdef BUILD_GRPC_FOR_BACKEND_LLAMA
 	PATH="${INSTALLED_PACKAGES}/bin:${PATH}" \
 	CMAKE_ARGS="${CMAKE_ARGS} ${ADDED_CMAKE_ARGS}" \
 	LLAMA_VERSION=$(CPPLLAMA_VERSION) \
-	$(MAKE) -C backend/cpp/${VARIANT} grpc-server
+	$(MAKE) -C backend/cpp/llama grpc-server
 else
 	echo "BUILD_GRPC_FOR_BACKEND_LLAMA is not defined."
-	LLAMA_VERSION=$(CPPLLAMA_VERSION) $(MAKE) -C backend/cpp/${VARIANT} grpc-server
+	LLAMA_VERSION=$(CPPLLAMA_VERSION) $(MAKE) -C backend/cpp/llama grpc-server
 endif
 
-backend-assets/grpc/llama-cpp-avx2: backend-assets/grpc
-	cp -rf backend/cpp/llama backend/cpp/llama-avx2
-	$(MAKE) -C backend/cpp/llama-avx2 purge
-	$(info ${GREEN}I llama-cpp build info:avx2${RESET})
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_AVX=on -DLLAMA_AVX2=on -DLLAMA_AVX512=off -DLLAMA_FMA=on -DLLAMA_F16C=on" $(MAKE) VARIANT="llama-avx2" build-llama-cpp-grpc-server
-	cp -rfv backend/cpp/llama-avx2/grpc-server backend-assets/grpc/llama-cpp-avx2
-
-backend-assets/grpc/llama-cpp-avx: backend-assets/grpc
-	cp -rf backend/cpp/llama backend/cpp/llama-avx
-	$(MAKE) -C backend/cpp/llama-avx purge
-	$(info ${GREEN}I llama-cpp build info:avx${RESET})
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_AVX=on -DLLAMA_AVX2=off -DLLAMA_AVX512=off -DLLAMA_FMA=off -DLLAMA_F16C=off" $(MAKE) VARIANT="llama-avx" build-llama-cpp-grpc-server
-	cp -rfv backend/cpp/llama-avx/grpc-server backend-assets/grpc/llama-cpp-avx
-
-backend-assets/grpc/llama-cpp-fallback: backend-assets/grpc
-	cp -rf backend/cpp/llama backend/cpp/llama-fallback
-	$(MAKE) -C backend/cpp/llama-fallback purge
-	$(info ${GREEN}I llama-cpp build info:fallback${RESET})
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_AVX=off -DLLAMA_AVX2=off -DLLAMA_AVX512=off -DLLAMA_FMA=off -DLLAMA_F16C=off" $(MAKE) VARIANT="llama-fallback" build-llama-cpp-grpc-server
-	cp -rfv backend/cpp/llama-fallback/grpc-server backend-assets/grpc/llama-cpp-fallback
+backend-assets/grpc/llama-cpp: backend-assets/grpc backend/cpp/llama/grpc-server
+	cp -rfv backend/cpp/llama/grpc-server backend-assets/grpc/llama-cpp
 # TODO: every binary should have its own folder instead, so can have different metal implementations
 ifeq ($(BUILD_TYPE),metal)
-	cp backend/cpp/llama-fallback/llama.cpp/build/bin/default.metallib backend-assets/grpc/
+	cp backend/cpp/llama/llama.cpp/build/bin/default.metallib backend-assets/grpc/
 endif
 
-backend-assets/grpc/llama-cpp-cuda: backend-assets/grpc
-	cp -rf backend/cpp/llama backend/cpp/llama-cuda
-	$(MAKE) -C backend/cpp/llama-cuda purge
-	$(info ${GREEN}I llama-cpp build info:cuda${RESET})
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_AVX=on -DLLAMA_AVX2=off -DLLAMA_AVX512=off -DLLAMA_FMA=off -DLLAMA_F16C=off -DLLAMA_CUDA=ON" $(MAKE) VARIANT="llama-cuda" build-llama-cpp-grpc-server
-	cp -rfv backend/cpp/llama-cuda/grpc-server backend-assets/grpc/llama-cpp-cuda
-
-backend-assets/grpc/llama-ggml: sources/go-llama.cpp sources/go-llama.cpp/libbinding.a backend-assets/grpc
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/go-llama.cpp LIBRARY_PATH=$(CURDIR)/sources/go-llama.cpp \
+backend-assets/grpc/llama-ggml: sources/go-llama-ggml sources/go-llama-ggml/libbinding.a backend-assets/grpc
+	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(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/
 
 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/
 
-backend-assets/grpc/rwkv: sources/go-rwkv.cpp sources/go-rwkv.cpp/librwkv.a backend-assets/grpc
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/go-rwkv.cpp LIBRARY_PATH=$(CURDIR)/sources/go-rwkv.cpp \
+backend-assets/grpc/rwkv: sources/go-rwkv sources/go-rwkv/librwkv.a backend-assets/grpc
+	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/go-rwkv LIBRARY_PATH=$(CURDIR)/sources/go-rwkv \
 	$(GOCMD) build -ldflags "$(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
@@ -733,7 +567,7 @@ docker:
 		--build-arg MAKEFLAGS="$(DOCKER_MAKEFLAGS)" \
 		--build-arg BUILD_TYPE=$(BUILD_TYPE) \
 		-t $(DOCKER_IMAGE) .
-
+	
 docker-aio:
 	@echo "Building AIO image with base $(BASE_IMAGE) as $(DOCKER_AIO_IMAGE)"
 	docker build \
@@ -747,7 +581,7 @@ docker-aio-all:
 
 docker-image-intel:
 	docker build \
-		--build-arg BASE_IMAGE=intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04 \
+		--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)" \
@@ -755,7 +589,7 @@ docker-image-intel:
 
 docker-image-intel-xpu:
 	docker build \
-		--build-arg BASE_IMAGE=intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04 \
+		--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)" \
@@ -763,4 +597,4 @@ docker-image-intel-xpu:
 
 .PHONY: swagger
 swagger:
-	swag init -g core/http/app.go --output swagger
+	swag init -g core/http/api.go --output swagger

README.md

@@ -44,24 +44,25 @@
 
 [![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. It is created and maintained by [Ettore Di Giacinto](https://github.com/mudler).
+**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.
 
 ## 🔥🔥 Hot topics / Roadmap
 
 [Roadmap](https://github.com/mudler/LocalAI/issues?q=is%3Aissue+is%3Aopen+label%3Aroadmap)
 
-- Chat, TTS, and Image generation in the WebUI: https://github.com/mudler/LocalAI/pull/2222
-- Reranker API: https://github.com/mudler/LocalAI/pull/2121
-- Gallery WebUI: https://github.com/mudler/LocalAI/pull/2104
-- llama3: https://github.com/mudler/LocalAI/discussions/2076
-- 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 / Tools API support: https://github.com/mudler/LocalAI/pull/1715
+- Upload file API: https://github.com/mudler/LocalAI/pull/1703
+- 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
 
 Hot topics (looking for contributors):
-
-- WebUI improvements: https://github.com/mudler/LocalAI/issues/2156
 - Backends v2: https://github.com/mudler/LocalAI/issues/1126
 - Improving UX v2: https://github.com/mudler/LocalAI/issues/1373
 - Assistant API: https://github.com/mudler/LocalAI/issues/1273
@@ -92,8 +93,7 @@ docker run -ti --name local-ai -p 8080:8080 localai/localai:latest-aio-cpu
 - 🧠 [Embeddings generation for vector databases](https://localai.io/features/embeddings/)
 - ✍️ [Constrained grammars](https://localai.io/features/constrained_grammars/)
 - 🖼️ [Download Models directly from Huggingface ](https://localai.io/models/)
-- 🥽 [Vision API](https://localai.io/features/gpt-vision/)
-- 🆕 [Reranker API](https://localai.io/features/reranker/)
+- 🆕 [Vision API](https://localai.io/features/gpt-vision/)
 
 ## 💻 Usage
 
@@ -114,7 +114,6 @@ Model galleries
 Other:
 - Helm chart https://github.com/go-skynet/helm-charts
 - VSCode extension https://github.com/badgooooor/localai-vscode-plugin
-- Terminal utility https://github.com/djcopley/ShellOracle
 - Local Smart assistant https://github.com/mudler/LocalAGI
 - Home Assistant https://github.com/sammcj/homeassistant-localai / https://github.com/drndos/hass-openai-custom-conversation
 - Discord bot https://github.com/mudler/LocalAGI/tree/main/examples/discord
@@ -133,7 +132,7 @@ Other:
 
 ## :book: 🎥 [Media, Blogs, Social](https://localai.io/basics/news/#media-blogs-social)
 
-- [Run LocalAI on AWS EKS with Pulumi](https://www.pulumi.com/blog/low-code-llm-apps-with-local-ai-flowise-and-pulumi/)
+- [Run LocalAI on AWS EKS with Pulumi](https://www.pulumi.com/ai/answers/tiZMDoZzZV6TLxgDXNBnFE/deploying-helm-charts-on-aws-eks)
 - [Run LocalAI on AWS](https://staleks.hashnode.dev/installing-localai-on-aws-ec2-instance)
 - [Create a slackbot for teams and OSS projects that answer to documentation](https://mudler.pm/posts/smart-slackbot-for-teams/)
 - [LocalAI meets k8sgpt](https://www.youtube.com/watch?v=PKrDNuJ_dfE)

aio/cpu/rerank.yaml

@@ -1,27 +0,0 @@
-name: jina-reranker-v1-base-en
-backend: rerankers
-parameters:
-  model: cross-encoder
-
-usage: |
-    You can test this model with curl like this:
-
-    curl http://localhost:8080/v1/rerank \
-      -H "Content-Type: application/json" \
-      -d '{
-      "model": "jina-reranker-v1-base-en",
-      "query": "Organic skincare products for sensitive skin",
-      "documents": [
-        "Eco-friendly kitchenware for modern homes",
-        "Biodegradable cleaning supplies for eco-conscious consumers",
-        "Organic cotton baby clothes for sensitive skin",
-        "Natural organic skincare range for sensitive skin",
-        "Tech gadgets for smart homes: 2024 edition",
-        "Sustainable gardening tools and compost solutions",
-        "Sensitive skin-friendly facial cleansers and toners",
-        "Organic food wraps and storage solutions",
-        "All-natural pet food for dogs with allergies",
-        "Yoga mats made from recycled materials"
-      ],
-      "top_n": 3
-    }'

aio/cpu/text-to-text.yaml

@@ -1,27 +1,20 @@
 name: gpt-4
 mmap: true
 parameters:
-  model: huggingface://NousResearch/Hermes-2-Pro-Llama-3-8B-GGUF/Hermes-2-Pro-Llama-3-8B-Q4_K_M.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 "tool"}}tool{{else if eq .RoleName "user"}}user{{end}}
-    {{- if .FunctionCall }}
-    <tool_call>
-    {{- else if eq .RoleName "tool" }}
-    <tool_response>
-    {{- end }}
+    {{- if .FunctionCall }}<tool_call>{{end}}
+    {{- if eq .RoleName "tool" }}<tool_result>{{end }}
     {{- if .Content}}
-    {{.Content }}
+    {{.Content}}
     {{- end }}
-    {{- if .FunctionCall}}
-    {{toJson .FunctionCall}}
-    {{- end }}
-    {{- if .FunctionCall }}
-    </tool_call>
-    {{- else if eq .RoleName "tool" }}
-    </tool_response>
-    {{- end }}<|im_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
@@ -36,7 +29,8 @@ template:
     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|>
+    </tool_call>
+    <|im_end|>
     {{.Input -}}
     <|im_start|>assistant
     <tool_call>

aio/entrypoint.sh

@@ -129,7 +129,7 @@ detect_gpu
 detect_gpu_size
 
 PROFILE="${PROFILE:-$GPU_SIZE}" # default to cpu
-export MODELS="${MODELS:-/aio/${PROFILE}/embeddings.yaml,/aio/${PROFILE}/rerank.yaml,/aio/${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}"
+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
 

aio/gpu-8g/rerank.yaml

@@ -1,27 +0,0 @@
-name: jina-reranker-v1-base-en
-backend: rerankers
-parameters:
-  model: cross-encoder
-
-usage: |
-    You can test this model with curl like this:
-
-    curl http://localhost:8080/v1/rerank \
-      -H "Content-Type: application/json" \
-      -d '{
-      "model": "jina-reranker-v1-base-en",
-      "query": "Organic skincare products for sensitive skin",
-      "documents": [
-        "Eco-friendly kitchenware for modern homes",
-        "Biodegradable cleaning supplies for eco-conscious consumers",
-        "Organic cotton baby clothes for sensitive skin",
-        "Natural organic skincare range for sensitive skin",
-        "Tech gadgets for smart homes: 2024 edition",
-        "Sustainable gardening tools and compost solutions",
-        "Sensitive skin-friendly facial cleansers and toners",
-        "Organic food wraps and storage solutions",
-        "All-natural pet food for dogs with allergies",
-        "Yoga mats made from recycled materials"
-      ],
-      "top_n": 3
-    }'

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

@@ -1,27 +1,20 @@
 name: gpt-4
 mmap: true
 parameters:
-  model: huggingface://NousResearch/Hermes-2-Pro-Llama-3-8B-GGUF/Hermes-2-Pro-Llama-3-8B-Q4_K_M.gguf
+  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>
-    {{- else if eq .RoleName "tool" }}
-    <tool_response>
-    {{- end }}
+    {{- if .FunctionCall }}<tool_call>{{end}}
+    {{- if eq .RoleName "tool" }}<tool_result>{{end }}
     {{- if .Content}}
-    {{.Content }}
+    {{.Content}}
     {{- end }}
-    {{- if .FunctionCall}}
-    {{toJson .FunctionCall}}
-    {{- end }}
-    {{- if .FunctionCall }}
-    </tool_call>
-    {{- else if eq .RoleName "tool" }}
-    </tool_response>
-    {{- end }}<|im_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
@@ -36,7 +29,8 @@ template:
     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|>
+    </tool_call>
+    <|im_end|>
     {{.Input -}}
     <|im_start|>assistant
     <tool_call>

aio/intel/rerank.yaml

@@ -1,27 +0,0 @@
-name: jina-reranker-v1-base-en
-backend: rerankers
-parameters:
-  model: cross-encoder
-
-usage: |
-    You can test this model with curl like this:
-
-    curl http://localhost:8080/v1/rerank \
-      -H "Content-Type: application/json" \
-      -d '{
-      "model": "jina-reranker-v1-base-en",
-      "query": "Organic skincare products for sensitive skin",
-      "documents": [
-        "Eco-friendly kitchenware for modern homes",
-        "Biodegradable cleaning supplies for eco-conscious consumers",
-        "Organic cotton baby clothes for sensitive skin",
-        "Natural organic skincare range for sensitive skin",
-        "Tech gadgets for smart homes: 2024 edition",
-        "Sustainable gardening tools and compost solutions",
-        "Sensitive skin-friendly facial cleansers and toners",
-        "Organic food wraps and storage solutions",
-        "All-natural pet food for dogs with allergies",
-        "Yoga mats made from recycled materials"
-      ],
-      "top_n": 3
-    }'

aio/intel/text-to-text.yaml

@@ -2,27 +2,20 @@ name: gpt-4
 mmap: false
 f16: false
 parameters:
-  model: huggingface://NousResearch/Hermes-2-Pro-Llama-3-8B-GGUF/Hermes-2-Pro-Llama-3-8B-Q4_K_M.gguf
+  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>
-    {{- else if eq .RoleName "tool" }}
-    <tool_response>
-    {{- end }}
+    {{- if .FunctionCall }}<tool_call>{{end}}
+    {{- if eq .RoleName "tool" }}<tool_result>{{end }}
     {{- if .Content}}
-    {{.Content }}
+    {{.Content}}
     {{- end }}
-    {{- if .FunctionCall}}
-    {{toJson .FunctionCall}}
-    {{- end }}
-    {{- if .FunctionCall }}
-    </tool_call>
-    {{- else if eq .RoleName "tool" }}
-    </tool_response>
-    {{- end }}<|im_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
@@ -37,7 +30,8 @@ template:
     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|>
+    </tool_call>
+    <|im_end|>
     {{.Input -}}
     <|im_start|>assistant
     <tool_call>

backend/backend.proto

@@ -23,30 +23,6 @@ service Backend {
   rpc StoresDelete(StoresDeleteOptions) returns (Result) {}
   rpc StoresGet(StoresGetOptions) returns (StoresGetResult) {}
   rpc StoresFind(StoresFindOptions) returns (StoresFindResult) {}
-
-  rpc Rerank(RerankRequest) returns (RerankResult) {}
-}
-
-message RerankRequest {
-  string query = 1;
-  repeated string documents = 2;
-  int32 top_n = 3;
-}
-
-message RerankResult {
-  Usage usage = 1;
-  repeated DocumentResult results = 2;
-}
-
-message Usage {
-  int32 total_tokens = 1;
-  int32 prompt_tokens = 2;
-}
-
-message DocumentResult {
-  int32 index = 1;
-  string text = 2;
-  float relevance_score = 3;
 }
 
 message StoresKey {
@@ -131,15 +107,11 @@ 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 {
@@ -201,7 +173,6 @@ message ModelOptions {
   bool   EnforceEager = 52;
   int32  SwapSpace = 53;
   int32  MaxModelLen = 54;
-  int32  TensorParallelSize = 55;
 
   string MMProj = 41;
 
@@ -212,9 +183,6 @@ message ModelOptions {
   float YarnBetaSlow = 47;
 
   string Type = 49;
-
-  bool FlashAttention = 56;
-  bool NoKVOffload = 57;
 }
 
 message Result {
@@ -288,8 +256,3 @@ message StatusResponse {
   State state = 1;
   MemoryUsageData memory = 2;
 }
-
-message Message {
-  string role = 1;
-  string content = 2;
-}

backend/backend_grpc.pb.go

@@ -0,0 +1,457 @@
+// 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

@@ -5,6 +5,7 @@ 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
@@ -51,7 +52,7 @@ $(GRPC_REPO):
 
 $(GRPC_BUILD): $(GRPC_REPO)
 	mkdir -p $(GRPC_BUILD)
-	cd $(GRPC_BUILD) && cmake $(CMAKE_ARGS) ../$(GRPC_REPO)/grpc && cmake --build . && cmake --build . --target install
+	cd $(GRPC_BUILD) && cmake $(CMAKE_ARGS) ../$(GRPC_REPO)/grpc && cmake --build . -- -j ${NUM_BUILD_THREADS} && cmake --build . --target install -- -j ${NUM_BUILD_THREADS}
 
 build: $(INSTALLED_PACKAGES)
 

backend/cpp/llama/Makefile

@@ -43,27 +43,35 @@ llama.cpp:
 
 llama.cpp/examples/grpc-server: llama.cpp
 	mkdir -p llama.cpp/examples/grpc-server
-	bash prepare.sh
+	cp -r $(abspath ./)/CMakeLists.txt llama.cpp/examples/grpc-server/
+	cp -r $(abspath ./)/grpc-server.cpp llama.cpp/examples/grpc-server/
+	cp -rfv $(abspath ./)/json.hpp llama.cpp/examples/grpc-server/
+	cp -rfv $(abspath ./)/utils.hpp llama.cpp/examples/grpc-server/
+	echo "add_subdirectory(grpc-server)" >> llama.cpp/examples/CMakeLists.txt
+## XXX: In some versions of CMake clip wasn't being built before llama.
+## This is an hack for now, but it should be fixed in the future.
+	cp -rfv llama.cpp/examples/llava/clip.h llama.cpp/examples/grpc-server/clip.h
+	cp -rfv llama.cpp/examples/llava/llava.cpp llama.cpp/examples/grpc-server/llava.cpp
+	echo '#include "llama.h"' > llama.cpp/examples/grpc-server/llava.h
+	cat llama.cpp/examples/llava/llava.h >> llama.cpp/examples/grpc-server/llava.h
+	cp -rfv llama.cpp/examples/llava/clip.cpp llama.cpp/examples/grpc-server/clip.cpp
 
 rebuild:
-	bash prepare.sh
+	cp -rfv $(abspath ./)/CMakeLists.txt llama.cpp/examples/grpc-server/
+	cp -rfv $(abspath ./)/grpc-server.cpp llama.cpp/examples/grpc-server/
+	cp -rfv $(abspath ./)/json.hpp llama.cpp/examples/grpc-server/
 	rm -rf grpc-server
 	$(MAKE) grpc-server
 
-purge:
-	rm -rf llama.cpp/build
-	rm -rf llama.cpp/examples/grpc-server
+clean:
+	rm -rf llama.cpp
 	rm -rf grpc-server
 
-clean: purge
-	rm -rf llama.cpp
-
 grpc-server: llama.cpp llama.cpp/examples/grpc-server
-	@echo "Building grpc-server with $(BUILD_TYPE) build type and $(CMAKE_ARGS)"
 ifneq (,$(findstring sycl,$(BUILD_TYPE)))
 	bash -c "source $(ONEAPI_VARS); \
-	cd llama.cpp && mkdir -p build && cd build && cmake .. $(CMAKE_ARGS) && $(MAKE)"
+	cd llama.cpp && mkdir -p build && cd build && cmake .. $(CMAKE_ARGS) && cmake --build . --config Release"	
 else
-	cd llama.cpp && mkdir -p build && cd build && cmake .. $(CMAKE_ARGS) && $(MAKE)
+	cd llama.cpp && mkdir -p build && cd build && cmake .. $(CMAKE_ARGS) && cmake --build . --config Release
 endif
 	cp llama.cpp/build/bin/grpc-server .

backend/cpp/llama/grpc-server.cpp

@@ -2254,9 +2254,6 @@ static void params_parse(const backend::ModelOptions* request,
     }
     params.use_mlock = request->mlock();
     params.use_mmap = request->mmap();
-    params.flash_attn = request->flashattention();
-    params.no_kv_offload = request->nokvoffload();
-
     params.embedding = request->embeddings();
 
     if (request->ropescaling() == "none")   { params.rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_NONE; }
@@ -2335,10 +2332,6 @@ 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);
@@ -2364,10 +2357,6 @@ 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/cpp/llama/prepare.sh

@@ -1,20 +0,0 @@
-#!/bin/bash
-
-cp -r CMakeLists.txt llama.cpp/examples/grpc-server/
-cp -r grpc-server.cpp llama.cpp/examples/grpc-server/
-cp -rfv json.hpp llama.cpp/examples/grpc-server/
-cp -rfv utils.hpp llama.cpp/examples/grpc-server/
-    
-if grep -q "grpc-server" llama.cpp/examples/CMakeLists.txt; then
-    echo "grpc-server already added"
-else
-    echo "add_subdirectory(grpc-server)" >> llama.cpp/examples/CMakeLists.txt
-fi
-
-## XXX: In some versions of CMake clip wasn't being built before llama.
-## This is an hack for now, but it should be fixed in the future.
-cp -rfv llama.cpp/examples/llava/clip.h llama.cpp/examples/grpc-server/clip.h
-cp -rfv llama.cpp/examples/llava/llava.cpp llama.cpp/examples/grpc-server/llava.cpp
-echo '#include "llama.h"' > llama.cpp/examples/grpc-server/llava.h
-cat llama.cpp/examples/llava/llava.h >> llama.cpp/examples/grpc-server/llava.h
-cp -rfv llama.cpp/examples/llava/clip.cpp llama.cpp/examples/grpc-server/clip.cpp

backend/go/llm/langchain/langchain.go

@@ -4,7 +4,6 @@ package main
 // It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
 import (
 	"fmt"
-	"os"
 
 	"github.com/go-skynet/LocalAI/pkg/grpc/base"
 	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
@@ -19,14 +18,9 @@ type LLM struct {
 }
 
 func (llm *LLM) Load(opts *pb.ModelOptions) error {
-	var err error
-	hfToken := os.Getenv("HUGGINGFACEHUB_API_TOKEN")
-	if hfToken == "" {
-		return fmt.Errorf("no huggingface token provided")
-	}
-	llm.langchain, err = langchain.NewHuggingFace(opts.Model, hfToken)
+	llm.langchain, _ = langchain.NewHuggingFace(opts.Model)
 	llm.model = opts.Model
-	return err
+	return nil
 }
 
 func (llm *LLM) Predict(opts *pb.PredictOptions) (string, error) {

backend/go/transcribe/transcript.go

@@ -11,8 +11,8 @@ import (
 	"github.com/go-skynet/LocalAI/core/schema"
 )
 
-func ffmpegCommand(args []string) (string, error) {
-	cmd := exec.Command("ffmpeg", args...) // Constrain this to ffmpeg to permit security scanner to see that the command is safe.
+func runCommand(command []string) (string, error) {
+	cmd := exec.Command(command[0], command[1:]...)
 	cmd.Env = os.Environ()
 	out, err := cmd.CombinedOutput()
 	return string(out), err
@@ -21,16 +21,16 @@ func ffmpegCommand(args []string) (string, error) {
 // AudioToWav converts audio to wav for transcribe.
 // TODO: use https://github.com/mccoyst/ogg?
 func audioToWav(src, dst string) error {
-	commandArgs := []string{"-i", src, "-format", "s16le", "-ar", "16000", "-ac", "1", "-acodec", "pcm_s16le", dst}
-	out, err := ffmpegCommand(commandArgs)
+    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)
 	}
 	return nil
 }
 
-func Transcript(model whisper.Model, audiopath, language string, threads uint) (schema.TranscriptionResult, error) {
-	res := schema.TranscriptionResult{}
+func Transcript(model whisper.Model, audiopath, language string, threads uint) (schema.Result, error) {
+	res := schema.Result{}
 
 	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.TranscriptionResult, error) {
+func (sd *Whisper) AudioTranscription(opts *pb.TranscriptRequest) (schema.Result, error) {
 	return Transcript(sd.whisper, opts.Dst, opts.Language, uint(opts.Threads))
 }

backend/python/autogptq/Makefile

@@ -1,17 +1,4 @@
 .PHONY: autogptq
-autogptq: protogen
-	bash install.sh
+autogptq:
+	$(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
-
-.PHONY: clean
-clean: protogen-clean
-	rm -rf venv __pycache__

backend/python/autogptq/autogptq.yml

@@ -0,0 +1,93 @@
+####
+# Attention! This file is abandoned. 
+# Please use the ../common-env/transformers/transformers.yml file to manage dependencies.
+###
+name: autogptq
+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
+  - wheel=0.41.2=py311h06a4308_0
+  - xz=5.4.2=h5eee18b_0
+  - zlib=1.2.13=h5eee18b_0
+  - pip:
+      - accelerate==0.27.0
+      - aiohttp==3.8.5
+      - aiosignal==1.3.1
+      - async-timeout==4.0.3
+      - attrs==23.1.0
+      - auto-gptq==0.7.1
+      - certifi==2023.7.22
+      - charset-normalizer==3.3.0
+      - datasets==2.14.5
+      - dill==0.3.7
+      - filelock==3.12.4
+      - frozenlist==1.4.0
+      - fsspec==2023.6.0
+      - grpcio==1.59.0
+      - huggingface-hub==0.16.4
+      - idna==3.4
+      - jinja2==3.1.2
+      - markupsafe==2.1.3
+      - mpmath==1.3.0
+      - multidict==6.0.4
+      - multiprocess==0.70.15
+      - networkx==3.1
+      - 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
+      - optimum==1.17.1
+      - packaging==23.2
+      - pandas==2.1.1
+      - peft==0.5.0
+      - protobuf==4.24.4
+      - psutil==5.9.5
+      - pyarrow==13.0.0
+      - python-dateutil==2.8.2
+      - pytz==2023.3.post1
+      - pyyaml==6.0.1
+      - regex==2023.10.3
+      - requests==2.31.0
+      - rouge==1.0.1
+      - safetensors>=0.3.3
+      - six==1.16.0
+      - sympy==1.12
+      - tokenizers==0.14.0
+      - 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
+      - urllib3==2.0.6
+      - xxhash==3.4.1
+      - yarl==1.9.2

backend/python/autogptq/backend_pb2_grpc.py

@@ -0,0 +1,363 @@
+# 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/autogptq/install.sh

@@ -1,14 +0,0 @@
-#!/bin/bash
-set -e
-
-source $(dirname $0)/../common/libbackend.sh
-
-# This is here because the Intel pip index is broken and returns 200 status codes for every package name, it just doesn't return any package links.
-# This makes uv think that the package exists in the Intel pip index, and by default it stops looking at other pip indexes once it finds a match.
-# We need uv to continue falling through to the pypi default index to find optimum[openvino] in the pypi index
-# the --upgrade actually allows us to *downgrade* torch to the version provided in the Intel pip index
-if [ "x${BUILD_PROFILE}" == "xintel" ]; then
-    EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
-fi
-
-installRequirements

backend/python/autogptq/requirements-intel.txt

@@ -1,4 +0,0 @@
---extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
-intel-extension-for-pytorch
-torch
-optimum[openvino]

backend/python/autogptq/requirements.txt

@@ -1,7 +0,0 @@
-accelerate
-auto-gptq==0.7.1
-grpcio==1.63.0
-protobuf
-torch
-certifi
-transformers

backend/python/autogptq/run.sh

@@ -1,4 +1,14 @@
 #!/bin/bash
-source $(dirname $0)/../common/libbackend.sh
 
-startBackend $@
+##
+## A bash script wrapper that runs the autogptq server with conda
+
+export PATH=$PATH:/opt/conda/bin
+
+# Activate conda environment
+source activate transformers
+
+# get the directory where the bash script is located
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
+
+python $DIR/autogptq.py $@

backend/python/autogptq/test.sh

@@ -1,6 +0,0 @@
-#!/bin/bash
-set -e
-
-source $(dirname $0)/../common/libbackend.sh
-
-runUnittests

backend/python/bark/Makefile

@@ -1,29 +1,15 @@
 .PHONY: ttsbark
-ttsbark: protogen
-	bash install.sh
+ttsbark:
+	$(MAKE) -C ../common-env/transformers
 
 .PHONY: run
-run: protogen
+run:
 	@echo "Running bark..."
 	bash run.sh
 	@echo "bark run."
 
 .PHONY: test
-test: protogen
+test:
 	@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
-
-.PHONY: clean
-clean: protogen-clean
-	rm -rf venv __pycache__

backend/python/bark/backend_pb2_grpc.py

@@ -0,0 +1,363 @@
+# 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/install.sh

@@ -1,14 +0,0 @@
-#!/bin/bash
-set -e
-
-source $(dirname $0)/../common/libbackend.sh
-
-# This is here because the Intel pip index is broken and returns 200 status codes for every package name, it just doesn't return any package links.
-# This makes uv think that the package exists in the Intel pip index, and by default it stops looking at other pip indexes once it finds a match.
-# We need uv to continue falling through to the pypi default index to find optimum[openvino] in the pypi index
-# the --upgrade actually allows us to *downgrade* torch to the version provided in the Intel pip index
-if [ "x${BUILD_PROFILE}" == "xintel" ]; then
-    EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
-fi
-
-installRequirements

backend/python/bark/requirements-intel.txt

@@ -1,5 +0,0 @@
---extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
-intel-extension-for-pytorch
-torch
-torchaudio
-optimum[openvino]

backend/python/bark/requirements.txt

@@ -1,6 +0,0 @@
-accelerate
-bark==0.1.5
-grpcio==1.63.0
-protobuf
-certifi
-transformers

backend/python/bark/run.sh

@@ -1,4 +1,14 @@
 #!/bin/bash
-source $(dirname $0)/../common/libbackend.sh
 
-startBackend $@
+##
+## A bash script wrapper that runs the ttsbark server with conda
+
+export PATH=$PATH:/opt/conda/bin
+
+# Activate conda environment
+source activate transformers
+
+# get the directory where the bash script is located
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
+
+python $DIR/ttsbark.py $@

backend/python/bark/test.py

@@ -18,7 +18,7 @@ class TestBackendServicer(unittest.TestCase):
         """
         This method sets up the gRPC service by starting the server
         """
-        self.service = subprocess.Popen(["python3", "backend.py", "--addr", "localhost:50051"])
+        self.service = subprocess.Popen(["python3", "ttsbark.py", "--addr", "localhost:50051"])
         time.sleep(10)
 
     def tearDown(self) -> None:

backend/python/bark/test.sh

@@ -1,6 +1,11 @@
 #!/bin/bash
-set -e
+##
+## A bash script wrapper that runs the bark server with conda
 
-source $(dirname $0)/../common/libbackend.sh
+# Activate conda environment
+source activate transformers
 
-runUnittests
+# 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.py

backend/python/common-env/transformers/Makefile

@@ -0,0 +1,21 @@
+CONDA_ENV_PATH = "transformers.yml"
+
+ifeq ($(BUILD_TYPE), cublas)
+	CONDA_ENV_PATH = "transformers-nvidia.yml"
+endif
+
+ifeq ($(BUILD_TYPE), hipblas)
+	CONDA_ENV_PATH = "transformers-rocm.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: transformers
+transformers:
+	@echo "Installing $(CONDA_ENV_PATH)..."
+	bash install.sh $(CONDA_ENV_PATH)

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

@@ -0,0 +1,38 @@
+#!/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 "transformers" ; then
+        echo "Creating virtual environment..."
+        conda env create --name transformers --file $1
+        echo "Virtual environment created."
+    else 
+        echo "Virtual environment already exists."
+    fi
+fi
+
+if [ -d "/opt/intel" ]; then
+    # Intel GPU: If the directory exists, we assume we are using the intel image
+    # (no conda env)
+    # https://github.com/intel/intel-extension-for-pytorch/issues/538
+    pip install intel-extension-for-transformers datasets sentencepiece tiktoken neural_speed optimum[openvino]
+fi
+
+if [ "$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

@@ -0,0 +1,123 @@
+name: transformers
+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
+  - wheel=0.41.2=py311h06a4308_0
+  - xz=5.4.2=h5eee18b_0
+  - zlib=1.2.13=h5eee18b_0
+  - pip:
+      - 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
+      - boto3==1.28.61
+      - botocore==1.31.61
+      - certifi==2023.7.22
+      - TTS==0.22.0
+      - charset-normalizer==3.3.0
+      - datasets==2.14.5
+      - sentence-transformers==2.5.1 # Updated Version
+      - sentencepiece==0.1.99
+      - dill==0.3.7
+      - einops==0.7.0
+      - encodec==0.1.1
+      - filelock==3.12.4
+      - frozenlist==1.4.0
+      - fsspec==2023.6.0
+      - funcy==2.0
+      - grpcio==1.59.0
+      - huggingface-hub
+      - idna==3.4
+      - jinja2==3.1.2
+      - jmespath==1.0.1
+      - markupsafe==2.1.3
+      - mpmath==1.3.0
+      - multidict==6.0.4
+      - multiprocess==0.70.15
+      - networkx
+      - 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
+      - optimum==1.17.1
+      - packaging==23.2
+      - pandas
+      - peft==0.5.0
+      - protobuf==4.24.4
+      - psutil==5.9.5
+      - pyarrow==13.0.0
+      - python-dateutil==2.8.2
+      - pytz==2023.3.post1
+      - pyyaml==6.0.1
+      - regex==2023.10.3
+      - requests==2.31.0
+      - rouge==1.0.1
+      - s3transfer==0.7.0
+      - safetensors>=0.4.1
+      - scipy==1.12.0 # Updated Version
+      - six==1.16.0
+      - 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
+      - urllib3==1.26.17
+      - xxhash==3.4.1
+      - yarl==1.9.2
+      - soundfile
+      - langid
+      - wget
+      - unidecode
+      - pyopenjtalk-prebuilt
+      - pypinyin
+      - inflect
+      - cn2an
+      - jieba
+      - eng_to_ipa
+      - openai-whisper
+      - matplotlib
+      - gradio==3.41.2
+      - nltk
+      - sudachipy
+      - sudachidict_core
+      - vocos
+      - vllm==0.3.2
+      - 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

@@ -0,0 +1,111 @@
+name: transformers
+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
+  - wheel=0.41.2=py311h06a4308_0
+  - xz=5.4.2=h5eee18b_0
+  - zlib=1.2.13=h5eee18b_0
+  - pip:
+      - --pre
+      - --extra-index-url https://download.pytorch.org/whl/nightly/
+      - accelerate==0.27.0
+      - auto-gptq==0.7.1
+      - aiohttp==3.8.5
+      - aiosignal==1.3.1
+      - 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
+      - TTS==0.22.0
+      - charset-normalizer==3.3.0
+      - datasets==2.14.5
+      - sentence-transformers==2.5.1 # Updated Version
+      - sentencepiece==0.1.99
+      - dill==0.3.7
+      - einops==0.7.0
+      - encodec==0.1.1
+      - filelock==3.12.4
+      - frozenlist==1.4.0
+      - fsspec==2023.6.0
+      - funcy==2.0
+      - grpcio==1.59.0
+      - huggingface-hub
+      - idna==3.4
+      - jinja2==3.1.2
+      - jmespath==1.0.1
+      - markupsafe==2.1.3
+      - mpmath==1.3.0
+      - multidict==6.0.4
+      - multiprocess==0.70.15
+      - networkx
+      - numpy==1.26.0
+      - packaging==23.2
+      - pandas
+      - peft==0.5.0
+      - protobuf==4.24.4
+      - psutil==5.9.5
+      - pyarrow==13.0.0
+      - python-dateutil==2.8.2
+      - pytz==2023.3.post1
+      - pyyaml==6.0.1
+      - regex==2023.10.3
+      - requests==2.31.0
+      - rouge==1.0.1
+      - s3transfer==0.7.0
+      - safetensors>=0.4.1
+      - scipy==1.12.0 # Updated Version
+      - six==1.16.0
+      - sympy==1.12
+      - tokenizers
+      - torch
+      - torchaudio
+      - tqdm==4.66.1
+      - triton==2.1.0
+      - typing-extensions==4.8.0
+      - tzdata==2023.3
+      - urllib3==1.26.17
+      - xxhash==3.4.1
+      - yarl==1.9.2
+      - soundfile
+      - langid
+      - wget
+      - unidecode
+      - optimum==1.17.1
+      - pyopenjtalk-prebuilt
+      - pypinyin
+      - inflect
+      - cn2an
+      - jieba
+      - eng_to_ipa
+      - openai-whisper
+      - matplotlib
+      - gradio==3.41.2
+      - nltk
+      - sudachipy
+      - sudachidict_core
+      - vocos
+      - vllm==0.3.2
+      - 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

@@ -0,0 +1,115 @@
+name: transformers
+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
+  - wheel=0.41.2=py311h06a4308_0
+  - xz=5.4.2=h5eee18b_0
+  - zlib=1.2.13=h5eee18b_0
+  - pip:
+      - 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
+      - sentence-transformers==2.5.1 # Updated Version
+      - sentencepiece==0.1.99
+      - dill==0.3.7
+      - einops==0.7.0
+      - encodec==0.1.1
+      - filelock==3.12.4
+      - frozenlist==1.4.0
+      - fsspec==2023.6.0
+      - funcy==2.0
+      - grpcio==1.59.0
+      - huggingface-hub
+      - humanfriendly==10.0
+      - idna==3.4
+      - jinja2==3.1.2
+      - jmespath==1.0.1
+      - markupsafe==2.1.3
+      - mpmath==1.3.0
+      - multidict==6.0.4
+      - 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
+      - protobuf==4.24.4
+      - psutil==5.9.5
+      - pyarrow==13.0.0
+      - python-dateutil==2.8.2
+      - pytz==2023.3.post1
+      - pyyaml==6.0.1
+      - regex==2023.10.3
+      - requests==2.31.0
+      - rouge==1.0.1
+      - s3transfer==0.7.0
+      - safetensors>=0.4.1
+      - scipy==1.12.0 # Updated Version
+      - six==1.16.0
+      - 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
+      - urllib3==1.26.17
+      - xxhash==3.4.1
+      - yarl==1.9.2
+      - soundfile
+      - langid
+      - wget
+      - unidecode
+      - pyopenjtalk-prebuilt
+      - pypinyin
+      - inflect
+      - cn2an
+      - jieba
+      - eng_to_ipa
+      - openai-whisper
+      - matplotlib
+      - gradio==3.41.2
+      - nltk
+      - sudachipy
+      - sudachidict_core
+      - vocos
+      - vllm==0.3.2
+      - 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/libbackend.sh

@@ -1,213 +0,0 @@
-
-
-# init handles the setup of the library
-# 
-# use the library by adding the following line to a script:
-# source $(dirname $0)/../common/libbackend.sh
-#
-# If you want to limit what targets a backend can be used on, set the variable LIMIT_TARGETS to a
-# space separated list of valid targets BEFORE sourcing the library, for example to only allow a backend
-# to be used on CUDA and CPU backends:
-#
-# LIMIT_TARGETS="cublas cpu"
-# source $(dirname $0)/../common/libbackend.sh
-#
-# You can use any valid BUILD_TYPE or BUILD_PROFILE, if you need to limit a backend to CUDA 12 only:
-#
-# LIMIT_TARGETS="cublas12"
-# source $(dirname $0)/../common/libbackend.sh
-#
-function init() {
-    BACKEND_NAME=${PWD##*/}
-    MY_DIR=$(realpath `dirname $0`)
-    BUILD_PROFILE=$(getBuildProfile)
-
-    # If a backend has defined a list of valid build profiles...
-    if [ ! -z "${LIMIT_TARGETS}" ]; then
-        isValidTarget=$(checkTargets ${LIMIT_TARGETS})
-        if [ ${isValidTarget} != true ]; then
-            echo "${BACKEND_NAME} can only be used on the following targets: ${LIMIT_TARGETS}"
-            exit 0
-        fi
-    fi
-
-    echo "Initializing libbackend for ${BACKEND_NAME}"
-}
-
-# getBuildProfile will inspect the system to determine which build profile is appropriate:
-# returns one of the following:
-# - cublas11
-# - cublas12
-# - hipblas
-# - intel
-function getBuildProfile() {
-    # First check if we are a cublas build, and if so report the correct build profile
-    if [ x"${BUILD_TYPE}" == "xcublas" ]; then
-        if [ ! -z ${CUDA_MAJOR_VERSION} ]; then
-            # If we have been given a CUDA version, we trust it
-            echo ${BUILD_TYPE}${CUDA_MAJOR_VERSION}
-        else
-            # We don't know what version of cuda we are, so we report ourselves as a generic cublas
-            echo ${BUILD_TYPE}
-        fi
-        return 0
-    fi
-
-    # If /opt/intel exists, then we are doing an intel/ARC build
-    if [ -d "/opt/intel" ]; then
-        echo "intel"
-        return 0
-    fi
-
-    # If for any other values of BUILD_TYPE, we don't need any special handling/discovery
-    if [ ! -z ${BUILD_TYPE} ]; then
-        echo ${BUILD_TYPE}
-        return 0
-    fi
-
-    # If there is no BUILD_TYPE set at all, set a build-profile value of CPU, we aren't building for any GPU targets
-    echo "cpu"
-}
-
-# ensureVenv makes sure that the venv for the backend both exists, and is activated.
-#
-# This function is idempotent, so you can call it as many times as you want and it will
-# always result in an activated virtual environment
-function ensureVenv() {
-    if [ ! -d "${MY_DIR}/venv" ]; then
-        uv venv ${MY_DIR}/venv
-        echo "virtualenv created"
-    fi
-    
-    if [ "x${VIRTUAL_ENV}" != "x${MY_DIR}/venv" ]; then
-        source ${MY_DIR}/venv/bin/activate
-        echo "virtualenv activated"
-    fi
-
-    echo "activated virtualenv has been ensured"
-}
-
-# installRequirements looks for several requirements files and if they exist runs the install for them in order
-#
-#  - requirements-install.txt
-#  - requirements.txt
-#  - requirements-${BUILD_TYPE}.txt
-#  - requirements-${BUILD_PROFILE}.txt
-#
-# BUILD_PROFILE is a pore specific version of BUILD_TYPE, ex: cuda11 or cuda12
-# it can also include some options that we do not have BUILD_TYPES for, ex: intel
-#
-# NOTE: for BUILD_PROFILE==intel, this function does NOT automatically use the Intel python package index.
-# you may want to add the following line to a requirements-intel.txt if you use one:
-#
-# --index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
-#
-# If you need to add extra flags into the pip install command you can do so by setting the variable EXTRA_PIP_INSTALL_FLAGS
-# before calling installRequirements.  For example:
-#
-# source $(dirname $0)/../common/libbackend.sh
-# EXTRA_PIP_INSTALL_FLAGS="--no-build-isolation"
-# installRequirements
-function installRequirements() {
-    ensureVenv
-
-    # These are the requirements files we will attempt to install, in order
-    declare -a requirementFiles=(
-        "${MY_DIR}/requirements-install.txt"
-        "${MY_DIR}/requirements.txt"
-        "${MY_DIR}/requirements-${BUILD_TYPE}.txt"
-    )
-
-    if [ "x${BUILD_TYPE}" != "x${BUILD_PROFILE}" ]; then
-        requirementFiles+=("${MY_DIR}/requirements-${BUILD_PROFILE}.txt")
-    fi
-
-    for reqFile in ${requirementFiles[@]}; do
-        if [ -f ${reqFile} ]; then
-            echo "starting requirements install for ${reqFile}"
-            uv pip install ${EXTRA_PIP_INSTALL_FLAGS} --requirement ${reqFile}
-            echo "finished requirements install for ${reqFile}"
-        fi
-    done
-}
-
-# startBackend discovers and runs the backend GRPC server
-#
-# You can specify a specific backend file to execute by setting BACKEND_FILE before calling startBackend.
-# example:
-#
-# source ../common/libbackend.sh
-# BACKEND_FILE="${MY_DIR}/source/backend.py"
-# startBackend $@
-#
-# valid filenames for autodiscovered backend servers are:
-#  - server.py
-#  - backend.py
-#  - ${BACKEND_NAME}.py
-function startBackend() {
-    ensureVenv
-
-    if [ ! -z ${BACKEND_FILE} ]; then
-        python ${BACKEND_FILE} $@
-    elif [ -e "${MY_DIR}/server.py" ]; then
-        python ${MY_DIR}/server.py $@
-    elif [ -e "${MY_DIR}/backend.py" ]; then
-        python ${MY_DIR}/backend.py $@
-    elif [ -e "${MY_DIR}/${BACKEND_NAME}.py" ]; then
-        python ${MY_DIR}/${BACKEND_NAME}.py $@
-    fi
-}
-
-# runUnittests discovers and runs python unittests
-#
-# You can specify a specific test file to use by setting TEST_FILE before calling runUnittests.
-# example:
-#
-# source ../common/libbackend.sh
-# TEST_FILE="${MY_DIR}/source/test.py"
-# runUnittests $@
-#
-# be default a file named test.py in the backends directory will be used
-function runUnittests() {
-    ensureVenv
-
-    if [ ! -z ${TEST_FILE} ]; then
-        testDir=$(dirname `realpath ${TEST_FILE}`)
-        testFile=$(basename ${TEST_FILE})
-        pushd ${testDir}
-        python -m unittest ${testFile}
-        popd
-    elif [ -f "${MY_DIR}/test.py" ]; then
-        pushd ${MY_DIR}
-        python -m unittest test.py
-        popd
-    else
-        echo "no tests defined for ${BACKEND_NAME}"
-    fi
-}
-
-##################################################################################
-# Below here are helper functions not intended to be used outside of the library #
-##################################################################################
-
-# checkTargets determines if the current BUILD_TYPE or BUILD_PROFILE is in a list of valid targets
-function checkTargets() {
-    # Collect all provided targets into a variable and...
-    targets=$@
-    # ...convert it into an array
-    declare -a targets=($targets)
-
-    for target in ${targets[@]}; do
-        if [ "x${BUILD_TYPE}" == "x${target}" ]; then
-            echo true
-            return 0
-        fi
-        if [ "x${BUILD_PROFILE}" == "x${target}" ]; then
-            echo true
-            return 0
-        fi
-    done
-    echo false
-}
-
-init

backend/python/common/template/Makefile

@@ -1,19 +0,0 @@
-.DEFAULT_GOAL := install
-
-.PHONY: install
-install: protogen
-	bash install.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
-
-.PHONY: clean
-clean: protogen-clean
-	rm -rf venv __pycache__

backend/python/common/template/backend.py

@@ -1,4 +0,0 @@
-#!/usr/bin/env python3
-import grpc
-import backend_pb2
-import backend_pb2_grpc

backend/python/common/template/install.sh

@@ -1,14 +0,0 @@
-#!/bin/bash
-set -e
-
-source $(dirname $0)/../common/libbackend.sh
-
-# This is here because the Intel pip index is broken and returns 200 status codes for every package name, it just doesn't return any package links.
-# This makes uv think that the package exists in the Intel pip index, and by default it stops looking at other pip indexes once it finds a match.
-# We need uv to continue falling through to the pypi default index to find optimum[openvino] in the pypi index
-# the --upgrade actually allows us to *downgrade* torch to the version provided in the Intel pip index
-if [ "x${BUILD_PROFILE}" == "xintel" ]; then
-    EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
-fi
-
-installRequirements

backend/python/common/template/requirements-intel.txt

@@ -1,4 +0,0 @@
---extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
-intel-extension-for-pytorch
-torch
-optimum[openvino]

backend/python/common/template/requirements.txt

@@ -1,2 +0,0 @@
-grpcio==1.63.0
-protobuf

backend/python/common/template/run.sh

@@ -1,4 +0,0 @@
-#!/bin/bash
-source $(dirname $0)/../common/libbackend.sh
-
-startBackend $@

backend/python/common/template/test.sh

@@ -1,6 +0,0 @@
-#!/bin/bash
-set -e
-
-source $(dirname $0)/../common/libbackend.sh
-
-runUnittests

backend/python/coqui/Makefile

@@ -1,29 +1,15 @@
 .PHONY: coqui
-coqui: protogen
-	bash install.sh
+coqui:
+	$(MAKE) -C ../common-env/transformers
 
 .PHONY: run
-run: protogen
+run:
 	@echo "Running coqui..."
 	bash run.sh
 	@echo "coqui run."
 
 .PHONY: test
-test: protogen
+test:
 	@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
-
-.PHONY: clean
-clean: protogen-clean
-	rm -rf venv __pycache__

backend/python/coqui/backend_pb2_grpc.py

@@ -0,0 +1,363 @@
+# 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/coqui/install.sh

@@ -1,14 +0,0 @@
-#!/bin/bash
-set -e
-
-source $(dirname $0)/../common/libbackend.sh
-
-# This is here because the Intel pip index is broken and returns 200 status codes for every package name, it just doesn't return any package links.
-# This makes uv think that the package exists in the Intel pip index, and by default it stops looking at other pip indexes once it finds a match.
-# We need uv to continue falling through to the pypi default index to find optimum[openvino] in the pypi index
-# the --upgrade actually allows us to *downgrade* torch to the version provided in the Intel pip index
-if [ "x${BUILD_PROFILE}" == "xintel" ]; then
-    EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
-fi
-
-installRequirements

backend/python/coqui/requirements-intel.txt

@@ -1,5 +0,0 @@
---extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
-intel-extension-for-pytorch
-torch
-torchaudio
-optimum[openvino]

backend/python/coqui/requirements.txt

@@ -1,6 +0,0 @@
-accelerate
-TTS==0.22.0
-grpcio==1.63.0
-protobuf
-certifi
-transformers

backend/python/coqui/run.sh

@@ -1,4 +1,14 @@
 #!/bin/bash
-source $(dirname $0)/../common/libbackend.sh
 
-startBackend $@
+##
+## A bash script wrapper that runs the ttsbark server with conda
+
+export PATH=$PATH:/opt/conda/bin
+
+# Activate conda environment
+source activate transformers
+
+# get the directory where the bash script is located
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
+
+python $DIR/coqui_server.py $@

backend/python/coqui/test.py

@@ -18,7 +18,7 @@ class TestBackendServicer(unittest.TestCase):
         """
         This method sets up the gRPC service by starting the server
         """
-        self.service = subprocess.Popen(["python3", "backend.py", "--addr", "localhost:50051"])
+        self.service = subprocess.Popen(["python3", "coqui_server.py", "--addr", "localhost:50051"])
         time.sleep(10)
 
     def tearDown(self) -> None:

backend/python/coqui/test.sh

@@ -1,6 +1,11 @@
 #!/bin/bash
-set -e
+##
+## A bash script wrapper that runs the bark server with conda
 
-source $(dirname $0)/../common/libbackend.sh
+# Activate conda environment
+source activate transformers
 
-runUnittests
+# 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.py

backend/python/diffusers/Makefile

@@ -12,28 +12,15 @@ export SKIP_CONDA=1
 endif
 
 .PHONY: diffusers
-diffusers: protogen
-	bash install.sh
+diffusers:
+	@echo "Installing $(CONDA_ENV_PATH)..."
+	bash install.sh $(CONDA_ENV_PATH)
 
 .PHONY: run
-run: protogen
+run:
 	@echo "Running diffusers..."
 	bash run.sh
 	@echo "Diffusers run."
 
-test: protogen
+test:
 	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
-
-.PHONY: clean
-clean: protogen-clean
-	rm -rf venv __pycache__

backend/python/diffusers/backend_pb2_grpc.py

@@ -0,0 +1,363 @@
+# 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/diffusers-rocm.yml

@@ -0,0 +1,64 @@
+name: diffusers
+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:
+      - --pre
+      - --extra-index-url https://download.pytorch.org/whl/nightly/
+      - accelerate>=0.11.0
+      - certifi==2023.7.22
+      - charset-normalizer==3.3.0
+      - compel==2.0.2
+      - diffusers==0.24.0
+      - filelock==3.12.4
+      - fsspec==2023.9.2
+      - grpcio==1.59.0
+      - huggingface-hub>=0.19.4
+      - idna==3.4
+      - importlib-metadata==6.8.0
+      - jinja2==3.1.2
+      - markupsafe==2.1.3
+      - mpmath==1.3.0
+      - networkx==3.1
+      - numpy==1.26.0
+      - omegaconf
+      - packaging==23.2
+      - pillow==10.0.1
+      - protobuf==4.24.4
+      - psutil==5.9.5
+      - pyparsing==3.1.1
+      - pyyaml==6.0.1
+      - regex==2023.10.3
+      - requests==2.31.0
+      - safetensors==0.4.0
+      - sympy==1.12
+      - tqdm==4.66.1
+      - transformers>=4.25.1
+      - triton==2.1.0
+      - typing-extensions==4.8.0
+      - urllib3==2.0.6
+      - zipp==3.17.0
+      - torch
+prefix: /opt/conda/envs/diffusers

backend/python/diffusers/diffusers.yml

@@ -0,0 +1,74 @@
+name: diffusers
+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
+      - certifi==2023.7.22
+      - charset-normalizer==3.3.0
+      - compel==2.0.2
+      - diffusers==0.24.0
+      - filelock==3.12.4
+      - fsspec==2023.9.2
+      - grpcio==1.59.0
+      - huggingface-hub>=0.19.4
+      - idna==3.4
+      - importlib-metadata==6.8.0
+      - jinja2==3.1.2
+      - markupsafe==2.1.3
+      - mpmath==1.3.0
+      - networkx==3.1
+      - 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
+      - omegaconf
+      - packaging==23.2
+      - pillow==10.0.1
+      - protobuf==4.24.4
+      - psutil==5.9.5
+      - pyparsing==3.1.1
+      - pyyaml==6.0.1
+      - regex==2023.10.3
+      - requests==2.31.0
+      - safetensors==0.4.0
+      - sympy==1.12
+      - torch==2.1.0
+      - tqdm==4.66.1
+      - transformers>=4.25.1
+      - triton==2.1.0
+      - typing-extensions==4.8.0
+      - urllib3==2.0.6
+      - zipp==3.17.0
+prefix: /opt/conda/envs/diffusers

backend/python/diffusers/install.sh

@@ -1,14 +1,50 @@
 #!/bin/bash
-set -e
+set -ex
 
-source $(dirname $0)/../common/libbackend.sh
+SKIP_CONDA=${SKIP_CONDA:-0}
 
-# This is here because the Intel pip index is broken and returns 200 status codes for every package name, it just doesn't return any package links.
-# This makes uv think that the package exists in the Intel pip index, and by default it stops looking at other pip indexes once it finds a match.
-# We need uv to continue falling through to the pypi default index to find optimum[openvino] in the pypi index
-# the --upgrade actually allows us to *downgrade* torch to the version provided in the Intel pip index
-if [ "x${BUILD_PROFILE}" == "xintel" ]; then
-    EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
+# Check if environment exist
+conda_env_exists(){
+    ! conda list --name "${@}" >/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 "diffusers" ; then
+        echo "Creating virtual environment..."
+        conda env create --name diffusers --file $1
+        echo "Virtual environment created."
+    else 
+        echo "Virtual environment already exists."
+    fi
 fi
 
-installRequirements
+if [ -d "/opt/intel" ]; then
+    # Intel GPU: If the directory exists, we assume we are using the Intel image
+    # https://github.com/intel/intel-extension-for-pytorch/issues/538
+    pip install torch==2.1.0a0 \
+                torchvision==0.16.0a0 \
+                torchaudio==2.1.0a0 \
+                intel-extension-for-pytorch==2.1.10+xpu \
+                --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+    
+    pip install google-api-python-client \
+                grpcio \
+                grpcio-tools \
+                diffusers==0.24.0 \
+                transformers>=4.25.1 \
+                accelerate \
+                compel==2.0.2 \
+                Pillow
+fi
+
+if [ "$PIP_CACHE_PURGE" = true ] ; then
+    if [ $SKIP_CONDA -ne 1 ]; then
+        # Activate conda environment
+        source activate diffusers
+    fi
+
+    pip cache purge
+fi

backend/python/diffusers/requirements-intel.txt

@@ -1,5 +0,0 @@
---extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
-intel-extension-for-pytorch
-torch
-torchvision
-optimum[openvino]

backend/python/diffusers/requirements.txt

@@ -1,10 +0,0 @@
-accelerate
-compel
-diffusers
-grpcio==1.63.0
-opencv-python
-pillow
-protobuf
-torch
-transformers
-certifi

backend/python/diffusers/run.sh

@@ -1,4 +1,19 @@
 #!/bin/bash
-source $(dirname $0)/../common/libbackend.sh
 
-startBackend $@
+##
+## A bash script wrapper that runs the diffusers server with conda
+
+if [ -d "/opt/intel" ]; then
+    # Assumes we are using the Intel oneAPI container image
+    # https://github.com/intel/intel-extension-for-pytorch/issues/538
+    export XPU=1
+else
+    export PATH=$PATH:/opt/conda/bin
+    # Activate conda environment
+    source activate diffusers
+fi
+
+# get the directory where the bash script is located
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
+
+python $DIR/backend_diffusers.py $@

backend/python/diffusers/test.py

@@ -18,7 +18,7 @@ class TestBackendServicer(unittest.TestCase):
         """
         This method sets up the gRPC service by starting the server
         """
-        self.service = subprocess.Popen(["python3", "backend.py", "--addr", "localhost:50051"])
+        self.service = subprocess.Popen(["python3", "backend_diffusers.py", "--addr", "localhost:50051"])
 
     def tearDown(self) -> None:
         """

backend/python/diffusers/test.sh

@@ -1,6 +1,14 @@
 #!/bin/bash
-set -e
 
-source $(dirname $0)/../common/libbackend.sh
+##
+## A bash script wrapper that runs the diffusers server with conda
 
-runUnittests
+export PATH=$PATH:/opt/conda/bin
+
+# Activate conda environment
+source activate diffusers
+
+# get the directory where the bash script is located
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
+
+python -m unittest $DIR/test.py

backend/python/exllama/.gitignore

@@ -1 +0,0 @@
-source

backend/python/exllama/Makefile

@@ -1,25 +1,11 @@
 export CONDA_ENV_PATH = "exllama.yml"
 
 .PHONY: exllama
-exllama: protogen
+exllama:
 	bash install.sh ${CONDA_ENV_PATH}
 
 .PHONY: run
-run: protogen
+run:
 	@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
-
-.PHONY: clean
-clean: protogen-clean
-	$(RM) -r venv source __pycache__