⬆️ Update mudler/go-stable-diffusion (#1674)

Signed-off-by: GitHub <noreply@github.com>
Co-authored-by: mudler <mudler@users.noreply.github.com>

.github/ISSUE_TEMPLATE/bug_report.md

@@ -2,9 +2,7 @@
 name: Bug report
 about: Create a report to help us improve
 title: ''
-labels: bug
-assignees: mudler
-
+labels: bug, unconfirmed, up-for-grabs
 ---
 
 <!-- Thanks for helping us to improve LocalAI! We welcome all bug reports. Please fill out each area of the template so we can better help you. Comments like this will be hidden when you post but you can delete them if you wish. -->

.github/ISSUE_TEMPLATE/feature_request.md

@@ -2,9 +2,7 @@
 name: Feature request
 about: Suggest an idea for this project
 title: ''
-labels: enhancement
-assignees: mudler
-
+labels: enhancement, up-for-grabs
 ---
 
 <!-- Thanks for helping us to improve LocalAI! We welcome all feature requests. Please fill out each area of the template so we can better help you. Comments like this will be hidden when you post but you can delete them if you wish. -->

.github/workflows/image-pr.yml

@@ -21,6 +21,7 @@ jobs:
       cuda-minor-version: ${{ matrix.cuda-minor-version }}
       platforms: ${{ matrix.platforms }}
       runs-on: ${{ matrix.runs-on }}
+      base-image: ${{ matrix.base-image }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
       dockerPassword: ${{ secrets.DOCKERHUB_PASSWORD }}
@@ -39,6 +40,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'extras'
             runs-on: 'arc-runner-set'
+            base-image: "ubuntu:22.04"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -48,6 +50,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'extras'
             runs-on: 'arc-runner-set'
+            base-image: "ubuntu:22.04"
   core-image-build:
     uses: ./.github/workflows/image_build.yml
     with:
@@ -60,6 +63,7 @@ jobs:
       cuda-minor-version: ${{ matrix.cuda-minor-version }}
       platforms: ${{ matrix.platforms }}
       runs-on: ${{ matrix.runs-on }}
+      base-image: ${{ matrix.base-image }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
       dockerPassword: ${{ secrets.DOCKERHUB_PASSWORD }}
@@ -75,6 +79,15 @@ jobs:
             ffmpeg: 'true'
             image-type: 'core'
             runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:22.04"
+          - build-type: 'sycl_f16'
+            platforms: 'linux/amd64'
+            tag-latest: 'false'
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
+            tag-suffix: 'sycl-f16-ffmpeg-core'
+            ffmpeg: 'true'
+            image-type: 'core'
+            runs-on: 'arc-runner-set'
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -84,3 +97,4 @@ jobs:
             ffmpeg: 'true'
             image-type: 'core'
             runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:22.04"

.github/workflows/image.yml

@@ -25,6 +25,7 @@ jobs:
       cuda-minor-version: ${{ matrix.cuda-minor-version }}
       platforms: ${{ matrix.platforms }}
       runs-on: ${{ matrix.runs-on }}
+      base-image: ${{ matrix.base-image }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
       dockerPassword: ${{ secrets.DOCKERHUB_PASSWORD }}
@@ -44,6 +45,7 @@ jobs:
             ffmpeg: ''
             image-type: 'extras'
             runs-on: 'arc-runner-set'
+            base-image: "ubuntu:22.04"
           - build-type: ''
             platforms: 'linux/amd64'
             tag-latest: 'false'
@@ -51,6 +53,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'extras'
             runs-on: 'arc-runner-set'
+            base-image: "ubuntu:22.04"
           - build-type: 'cublas'
             cuda-major-version: "11"
             cuda-minor-version: "7"
@@ -60,6 +63,7 @@ jobs:
             ffmpeg: ''
             image-type: 'extras'
             runs-on: 'arc-runner-set'
+            base-image: "ubuntu:22.04"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -69,6 +73,7 @@ jobs:
             ffmpeg: ''
             image-type: 'extras'
             runs-on: 'arc-runner-set'
+            base-image: "ubuntu:22.04"
           - build-type: 'cublas'
             cuda-major-version: "11"
             cuda-minor-version: "7"
@@ -78,6 +83,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'extras'
             runs-on: 'arc-runner-set'
+            base-image: "ubuntu:22.04"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -87,6 +93,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'extras'
             runs-on: 'arc-runner-set'
+            base-image: "ubuntu:22.04"
           - build-type: ''
             #platforms: 'linux/amd64,linux/arm64'
             platforms: 'linux/amd64'
@@ -94,6 +101,7 @@ jobs:
             tag-suffix: ''
             ffmpeg: ''
             image-type: 'extras'
+            base-image: "ubuntu:22.04"
             runs-on: 'arc-runner-set'
   core-image-build:
     uses: ./.github/workflows/image_build.yml
@@ -107,6 +115,7 @@ jobs:
       cuda-minor-version: ${{ matrix.cuda-minor-version }}
       platforms: ${{ matrix.platforms }}
       runs-on: ${{ matrix.runs-on }}
+      base-image: ${{ matrix.base-image }}
     secrets:
       dockerUsername: ${{ secrets.DOCKERHUB_USERNAME }}
       dockerPassword: ${{ secrets.DOCKERHUB_PASSWORD }}
@@ -121,7 +130,40 @@ jobs:
             tag-suffix: '-ffmpeg-core'
             ffmpeg: 'true'
             image-type: 'core'
+            base-image: "ubuntu:22.04"
             runs-on: 'ubuntu-latest'
+          - build-type: 'sycl_f16'
+            platforms: 'linux/amd64'
+            tag-latest: 'false'
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
+            tag-suffix: '-sycl-f16-core'
+            ffmpeg: 'false'
+            image-type: 'core'
+            runs-on: 'arc-runner-set'
+          - build-type: 'sycl_f32'
+            platforms: 'linux/amd64'
+            tag-latest: 'false'
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
+            tag-suffix: '-sycl-f32-core'
+            ffmpeg: 'false'
+            image-type: 'core'
+            runs-on: 'arc-runner-set'
+          - build-type: 'sycl_f16'
+            platforms: 'linux/amd64'
+            tag-latest: 'false'
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
+            tag-suffix: '-sycl-f16-ffmpeg-core'
+            ffmpeg: 'true'
+            image-type: 'core'
+            runs-on: 'arc-runner-set'
+          - build-type: 'sycl_f32'
+            platforms: 'linux/amd64'
+            tag-latest: 'false'
+            base-image: "intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04"
+            tag-suffix: '-sycl-f32-ffmpeg-core'
+            ffmpeg: 'true'
+            image-type: 'core'
+            runs-on: 'arc-runner-set'
           - build-type: 'cublas'
             cuda-major-version: "11"
             cuda-minor-version: "7"
@@ -130,6 +172,7 @@ jobs:
             tag-suffix: '-cublas-cuda11-core'
             ffmpeg: ''
             image-type: 'core'
+            base-image: "ubuntu:22.04"
             runs-on: 'ubuntu-latest'
           - build-type: 'cublas'
             cuda-major-version: "12"
@@ -139,6 +182,7 @@ jobs:
             tag-suffix: '-cublas-cuda12-core'
             ffmpeg: ''
             image-type: 'core'
+            base-image: "ubuntu:22.04"
             runs-on: 'ubuntu-latest'
           - build-type: 'cublas'
             cuda-major-version: "11"
@@ -149,6 +193,7 @@ jobs:
             ffmpeg: 'true'
             image-type: 'core'
             runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:22.04"
           - build-type: 'cublas'
             cuda-major-version: "12"
             cuda-minor-version: "1"
@@ -158,3 +203,4 @@ jobs:
             ffmpeg: 'true'
             image-type: 'core'
             runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:22.04"

.github/workflows/image_build.yml

@@ -4,6 +4,11 @@ name: 'build container images (reusable)'
 on:
   workflow_call:
     inputs:
+      base-image:
+        description: 'Base image'
+        required: false
+        default: ''
+        type: string
       build-type:
         description: 'Build type'
         default: ''
@@ -64,42 +69,47 @@ jobs:
           && sudo apt-get install -y git
       - name: Checkout
         uses: actions/checkout@v4
-      # - name: Release space from worker
-      #   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
-      #     df -h
+      - name: Release space from worker
+        if: inputs.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 "$AGENT_TOOLSDIRECTORY" || true
+          df -h
       - name: Docker meta
         id: meta
         uses: docker/metadata-action@v5
@@ -149,6 +159,7 @@ jobs:
             CUDA_MINOR_VERSION=${{ inputs.cuda-minor-version }}
             FFMPEG=${{ inputs.ffmpeg }}
             IMAGE_TYPE=${{ inputs.image-type }}
+            BASE_IMAGE=${{ inputs.base-image }}
           context: .
           file: ./Dockerfile
           platforms: ${{ inputs.platforms }}

.github/workflows/test-extra.yml

@@ -164,34 +164,74 @@ jobs:
 
            
 
-  tests-bark:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Clone
-        uses: actions/checkout@v4
-        with: 
-          submodules: true
-      - name: Dependencies
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential ffmpeg
-          curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
-             sudo install -o root -g root -m 644 conda.gpg /usr/share/keyrings/conda-archive-keyring.gpg && \
-              gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \
-             sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" > /etc/apt/sources.list.d/conda.list' && \
-             sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
-             sudo apt-get update && \
-             sudo apt-get install -y conda
-          sudo apt-get install -y ca-certificates cmake curl patch
-          sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
+  # tests-bark:
+  #   runs-on: ubuntu-latest
+  #   steps:
+  #     - name: Release space from worker
+  #       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: Clone
+  #       uses: actions/checkout@v4
+  #       with: 
+  #         submodules: true
+  #     - name: Dependencies
+  #       run: |
+  #         sudo apt-get update
+  #         sudo apt-get install build-essential ffmpeg
+  #         curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
+  #            sudo install -o root -g root -m 644 conda.gpg /usr/share/keyrings/conda-archive-keyring.gpg && \
+  #             gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \
+  #            sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" > /etc/apt/sources.list.d/conda.list' && \
+  #            sudo /bin/bash -c 'echo "deb [arch=amd64 signed-by=/usr/share/keyrings/conda-archive-keyring.gpg] https://repo.anaconda.com/pkgs/misc/debrepo/conda stable main" | tee -a /etc/apt/sources.list.d/conda.list' && \
+  #            sudo apt-get update && \
+  #            sudo apt-get install -y conda
+  #         sudo apt-get install -y ca-certificates cmake curl patch
+  #         sudo apt-get install -y libopencv-dev && sudo ln -s /usr/include/opencv4/opencv2 /usr/include/opencv2
           
-          sudo rm -rfv /usr/bin/conda || true
+  #         sudo rm -rfv /usr/bin/conda || true
 
-      - name: Test bark
-        run: |
-           export PATH=$PATH:/opt/conda/bin
-           make -C backend/python/bark
-           make -C backend/python/bark test
+  #     - name: Test bark
+  #       run: |
+  #          export PATH=$PATH:/opt/conda/bin
+  #          make -C backend/python/bark
+  #          make -C backend/python/bark test
 
            
   # Below tests needs GPU. Commented out for now
@@ -274,4 +314,4 @@ jobs:
         run: |
            export PATH=$PATH:/opt/conda/bin
            make -C backend/python/coqui
-           make -C backend/python/coqui test
+           make -C backend/python/coqui test

.gitmodules

@@ -1,3 +1,6 @@
 [submodule "docs/themes/hugo-theme-relearn"]
 	path = docs/themes/hugo-theme-relearn
 	url = https://github.com/McShelby/hugo-theme-relearn.git
+[submodule "docs/themes/lotusdocs"]
+	path = docs/themes/lotusdocs
+	url = https://github.com/colinwilson/lotusdocs

Dockerfile

@@ -1,10 +1,11 @@
-ARG GO_VERSION=1.21-bullseye
+ARG GO_VERSION=1.21
 ARG IMAGE_TYPE=extras
+ARG BASE_IMAGE=ubuntu:22.04
+
 # extras or core
+FROM ${BASE_IMAGE} as requirements-core
 
-
-FROM golang:$GO_VERSION as requirements-core
-
+ARG GO_VERSION=1.21.7
 ARG BUILD_TYPE
 ARG CUDA_MAJOR_VERSION=11
 ARG CUDA_MINOR_VERSION=7
@@ -12,15 +13,17 @@ 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,autogptq:/build/backend/python/autogptq/run.sh,bark:/build/backend/python/bark/run.sh,diffusers:/build/backend/python/diffusers/run.sh,exllama:/build/backend/python/exllama/run.sh,vall-e-x:/build/backend/python/vall-e-x/run.sh,vllm:/build/backend/python/vllm/run.sh,mamba:/build/backend/python/mamba/run.sh,exllama2:/build/backend/python/exllama2/run.sh,transformers-musicgen:/build/backend/python/transformers-musicgen/run.sh"
 
-ENV EXTERNAL_GRPC_BACKENDS="coqui:/build/backend/python/coqui/run.sh,huggingface-embeddings:/build/backend/python/sentencetransformers/run.sh,petals:/build/backend/python/petals/run.sh,transformers:/build/backend/python/transformers/run.sh,sentencetransformers:/build/backend/python/sentencetransformers/run.sh,autogptq:/build/backend/python/autogptq/run.sh,bark:/build/backend/python/bark/run.sh,diffusers:/build/backend/python/diffusers/run.sh,exllama:/build/backend/python/exllama/run.sh,vall-e-x:/build/backend/python/vall-e-x/run.sh,vllm:/build/backend/python/vllm/run.sh,exllama2:/build/backend/python/exllama2/run.sh,transformers-musicgen:/build/backend/python/transformers-musicgen/run.sh"
-
-ENV GALLERIES='[{"name":"model-gallery", "url":"github:go-skynet/model-gallery/index.yaml"}, {"url": "github:go-skynet/model-gallery/huggingface.yaml","name":"huggingface"}]'
 ARG GO_TAGS="stablediffusion tinydream tts"
 
 RUN apt-get update && \
-    apt-get install -y ca-certificates curl patch pip cmake && apt-get clean
+    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 -v -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
@@ -32,13 +35,13 @@ RUN echo "Target Variant: $TARGETVARIANT"
 # CuBLAS requirements
 RUN if [ "${BUILD_TYPE}" = "cublas" ]; then \
     apt-get install -y software-properties-common && \
-    apt-add-repository contrib && \
-    curl -O https://developer.download.nvidia.com/compute/cuda/repos/debian11/x86_64/cuda-keyring_1.0-1_all.deb && \
-    dpkg -i cuda-keyring_1.0-1_all.deb && \
-    rm -f cuda-keyring_1.0-1_all.deb && \
+    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} 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
+
 ENV PATH /usr/local/cuda/bin:${PATH}
 
 # OpenBLAS requirements and stable diffusion
@@ -64,12 +67,12 @@ RUN curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmo
     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 install -y conda && apt-get clean
 
 ENV PATH="/root/.cargo/bin:${PATH}"
 RUN pip install --upgrade pip
 RUN curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
-RUN apt-get install -y espeak-ng espeak
+RUN apt-get install -y espeak-ng espeak && apt-get clean
 
 ###################################
 ###################################
@@ -127,10 +130,11 @@ 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 install -y ffmpeg \
+    apt-get install -y ffmpeg && apt-get clean \
     ; fi
 
 WORKDIR /build
@@ -168,6 +172,9 @@ RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
 RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
 	PATH=$PATH:/opt/conda/bin make -C backend/python/vllm \
     ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+	PATH=$PATH:/opt/conda/bin make -C backend/python/mamba \
+    ; fi
 RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
 	PATH=$PATH:/opt/conda/bin make -C backend/python/sentencetransformers \
     ; fi
@@ -193,6 +200,9 @@ RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
 	PATH=$PATH:/opt/conda/bin make -C backend/python/coqui \
     ; fi
 
+# Make sure the models directory exists
+RUN mkdir -p /build/models
+
 # Define the health check command
 HEALTHCHECK --interval=1m --timeout=10m --retries=10 \
   CMD curl -f $HEALTHCHECK_ENDPOINT || exit 1

LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2023 Ettore Di Giacinto
+Copyright (c) 2023-2024 Ettore Di Giacinto (mudler@localai.io)
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

Makefile

@@ -8,15 +8,12 @@ GOLLAMA_VERSION?=aeba71ee842819da681ea537e78846dc75949ac0
 
 GOLLAMA_STABLE_VERSION?=50cee7712066d9e38306eccadcfbb44ea87df4b7
 
-CPPLLAMA_VERSION?=eec22a1c6378d9a013943cbddb4330c0da621442
+CPPLLAMA_VERSION?=4b7b38bef5addbd31f453871d79647fbae6bec8a
 
 # gpt4all version
 GPT4ALL_REPO?=https://github.com/nomic-ai/gpt4all
 GPT4ALL_VERSION?=27a8b020c36b0df8f8b82a252d261cda47cf44b8
 
-# go-ggml-transformers version
-GOGGMLTRANSFORMERS_VERSION?=ffb09d7dd71e2cbc6c5d7d05357d230eea6f369a
-
 # go-rwkv version
 RWKV_REPO?=https://github.com/donomii/go-rwkv.cpp
 RWKV_VERSION?=633c5a3485c403cb2520693dc0991a25dace9f0f
@@ -31,7 +28,7 @@ BERT_VERSION?=6abe312cded14042f6b7c3cd8edf082713334a4d
 PIPER_VERSION?=d6b6275ba037dabdba4a8b65dfdf6b2a73a67f07
 
 # stablediffusion version
-STABLEDIFFUSION_VERSION?=902db5f066fd137697e3b69d0fa10d4782bd2c2f
+STABLEDIFFUSION_VERSION?=d5d2be8e7e395c2d73ceef61e6fe8d240f2cd831
 
 # tinydream version
 TINYDREAM_VERSION?=772a9c0d9aaf768290e63cca3c904fe69faf677a
@@ -140,12 +137,21 @@ endif
 ifeq ($(findstring tts,$(GO_TAGS)),tts)
 #	OPTIONAL_TARGETS+=go-piper/libpiper_binding.a
 #	OPTIONAL_TARGETS+=backend-assets/espeak-ng-data
-	PIPER_CGO_CXXFLAGS+=-I$(shell pwd)/sources/go-piper/piper/src/cpp -I$(shell pwd)/sources/go-piper/piper/build/fi/include -I$(shell pwd)/sources/go-piper/piper/build/pi/include -I$(shell pwd)/sources/go-piper/piper/build/si/include
- 	PIPER_CGO_LDFLAGS+=-L$(shell pwd)/sources/go-piper/piper/build/fi/lib -L$(shell pwd)/sources/go-piper/piper/build/pi/lib -L$(shell pwd)/sources/go-piper/piper/build/si/lib -lfmt -lspdlog -lucd
+	PIPER_CGO_CXXFLAGS+=-I$(CURDIR)/sources/go-piper/piper/src/cpp -I$(CURDIR)/sources/go-piper/piper/build/fi/include -I$(CURDIR)/sources/go-piper/piper/build/pi/include -I$(CURDIR)/sources/go-piper/piper/build/si/include
+	PIPER_CGO_LDFLAGS+=-L$(CURDIR)/sources/go-piper/piper/build/fi/lib -L$(CURDIR)/sources/go-piper/piper/build/pi/lib -L$(CURDIR)/sources/go-piper/piper/build/si/lib -lfmt -lspdlog -lucd
 	OPTIONAL_GRPC+=backend-assets/grpc/piper
 endif
 
-ALL_GRPC_BACKENDS=backend-assets/grpc/langchain-huggingface backend-assets/grpc/falcon-ggml backend-assets/grpc/bert-embeddings backend-assets/grpc/llama backend-assets/grpc/llama-cpp backend-assets/grpc/llama-ggml backend-assets/grpc/gpt4all backend-assets/grpc/dolly backend-assets/grpc/gpt2 backend-assets/grpc/gptj backend-assets/grpc/gptneox backend-assets/grpc/mpt backend-assets/grpc/replit backend-assets/grpc/starcoder backend-assets/grpc/rwkv backend-assets/grpc/whisper $(OPTIONAL_GRPC)
+ALL_GRPC_BACKENDS=backend-assets/grpc/langchain-huggingface
+ALL_GRPC_BACKENDS+=backend-assets/grpc/bert-embeddings
+ALL_GRPC_BACKENDS+=backend-assets/grpc/llama
+ALL_GRPC_BACKENDS+=backend-assets/grpc/llama-cpp
+ALL_GRPC_BACKENDS+=backend-assets/grpc/llama-ggml
+ALL_GRPC_BACKENDS+=backend-assets/grpc/gpt4all
+ALL_GRPC_BACKENDS+=backend-assets/grpc/rwkv
+ALL_GRPC_BACKENDS+=backend-assets/grpc/whisper
+ALL_GRPC_BACKENDS+=$(OPTIONAL_GRPC)
+
 GRPC_BACKENDS?=$(ALL_GRPC_BACKENDS) $(OPTIONAL_GRPC)
 
 # If empty, then we build all
@@ -153,6 +159,10 @@ ifeq ($(GRPC_BACKENDS),)
 	GRPC_BACKENDS=$(ALL_GRPC_BACKENDS)
 endif
 
+ifeq ($(BUILD_API_ONLY),true)
+	GRPC_BACKENDS=
+endif
+
 .PHONY: all test build vendor
 
 all: help
@@ -213,14 +223,6 @@ backend-assets/espeak-ng-data: sources/go-piper
 sources/gpt4all/gpt4all-bindings/golang/libgpt4all.a: sources/gpt4all
 	$(MAKE) -C sources/gpt4all/gpt4all-bindings/golang/ libgpt4all.a
 
-## CEREBRAS GPT
-sources/go-ggml-transformers:
-	git clone --recurse-submodules https://github.com/go-skynet/go-ggml-transformers.cpp sources/go-ggml-transformers
-	cd sources/go-ggml-transformers && git checkout -b build $(GOGPT2_VERSION) && git submodule update --init --recursive --depth 1
-
-sources/go-ggml-transformers/libtransformers.a: sources/go-ggml-transformers
-	$(MAKE) -C sources/go-ggml-transformers BUILD_TYPE=$(BUILD_TYPE) libtransformers.a
-
 sources/whisper.cpp:
 	git clone https://github.com/ggerganov/whisper.cpp.git sources/whisper.cpp
 	cd sources/whisper.cpp && git checkout -b build $(WHISPER_CPP_VERSION) && git submodule update --init --recursive --depth 1
@@ -248,19 +250,18 @@ sources/go-piper/libpiper_binding.a: sources/go-piper
 backend/cpp/llama/llama.cpp:
 	LLAMA_VERSION=$(CPPLLAMA_VERSION) $(MAKE) -C backend/cpp/llama llama.cpp	
 
-get-sources: backend/cpp/llama/llama.cpp sources/go-llama sources/go-llama-ggml sources/go-ggml-transformers sources/gpt4all sources/go-piper sources/go-rwkv sources/whisper.cpp sources/go-bert sources/go-stable-diffusion sources/go-tiny-dream
+get-sources: backend/cpp/llama/llama.cpp sources/go-llama 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
 	touch $@
 
 replace:
-	$(GOCMD) mod edit -replace github.com/nomic-ai/gpt4all/gpt4all-bindings/golang=$(shell pwd)/sources/gpt4all/gpt4all-bindings/golang
-	$(GOCMD) mod edit -replace github.com/go-skynet/go-ggml-transformers.cpp=$(shell pwd)/sources/go-ggml-transformers
-	$(GOCMD) mod edit -replace github.com/donomii/go-rwkv.cpp=$(shell pwd)/sources/go-rwkv
-	$(GOCMD) mod edit -replace github.com/ggerganov/whisper.cpp=$(shell pwd)/sources/whisper.cpp
-	$(GOCMD) mod edit -replace github.com/ggerganov/whisper.cpp/bindings/go=$(shell pwd)/sources/whisper.cpp/bindings/go
-	$(GOCMD) mod edit -replace github.com/go-skynet/go-bert.cpp=$(shell pwd)/sources/go-bert
-	$(GOCMD) mod edit -replace github.com/mudler/go-stable-diffusion=$(shell pwd)/sources/go-stable-diffusion
-	$(GOCMD) mod edit -replace github.com/M0Rf30/go-tiny-dream=$(shell pwd)/sources/go-tiny-dream
-	$(GOCMD) mod edit -replace github.com/mudler/go-piper=$(shell pwd)/sources/go-piper
+	$(GOCMD) mod edit -replace github.com/nomic-ai/gpt4all/gpt4all-bindings/golang=$(CURDIR)/sources/gpt4all/gpt4all-bindings/golang
+	$(GOCMD) mod edit -replace github.com/donomii/go-rwkv.cpp=$(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
+	$(GOCMD) mod edit -replace github.com/mudler/go-stable-diffusion=$(CURDIR)/sources/go-stable-diffusion
+	$(GOCMD) mod edit -replace github.com/M0Rf30/go-tiny-dream=$(CURDIR)/sources/go-tiny-dream
+	$(GOCMD) mod edit -replace github.com/mudler/go-piper=$(CURDIR)/sources/go-piper
 
 prepare-sources: get-sources replace
 	$(GOCMD) mod download
@@ -272,7 +273,6 @@ rebuild: ## Rebuilds the project
 	$(MAKE) -C sources/go-llama clean
 	$(MAKE) -C sources/go-llama-ggml clean
 	$(MAKE) -C sources/gpt4all/gpt4all-bindings/golang/ clean
-	$(MAKE) -C sources/go-ggml-transformers clean
 	$(MAKE) -C sources/go-rwkv clean
 	$(MAKE) -C sources/whisper.cpp clean
 	$(MAKE) -C sources/go-stable-diffusion clean
@@ -290,19 +290,17 @@ clean: ## Remove build related file
 	rm -rf ./sources
 	rm -rf $(BINARY_NAME)
 	rm -rf release/
-	rm -rf ./backend/cpp/grpc/grpc_repo
-	rm -rf ./backend/cpp/grpc/build
-	rm -rf ./backend/cpp/grpc/installed_packages
+	rm -rf backend-assets
+	$(MAKE) -C backend/cpp/grpc clean
 	$(MAKE) -C backend/cpp/llama clean
 
 ## Build:
 
-build: grpcs prepare ## Build the project
+build: backend-assets grpcs prepare ## Build the project
 	$(info ${GREEN}I local-ai build info:${RESET})
 	$(info ${GREEN}I BUILD_TYPE: ${YELLOW}$(BUILD_TYPE)${RESET})
 	$(info ${GREEN}I GO_TAGS: ${YELLOW}$(GO_TAGS)${RESET})
 	$(info ${GREEN}I LD_FLAGS: ${YELLOW}$(LD_FLAGS)${RESET})
-
 	CGO_LDFLAGS="$(CGO_LDFLAGS)" $(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o $(BINARY_NAME) ./
 
 dist: build
@@ -319,7 +317,7 @@ run: prepare ## run local-ai
 test-models/testmodel:
 	mkdir test-models
 	mkdir test-dir
-	wget -q https://huggingface.co/nnakasato/ggml-model-test/resolve/main/ggml-model-q4.bin -O test-models/testmodel
+	wget -q https://huggingface.co/TheBloke/orca_mini_3B-GGML/resolve/main/orca-mini-3b.ggmlv3.q4_0.bin -O test-models/testmodel
 	wget -q https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-base.en.bin -O test-models/whisper-en
 	wget -q https://huggingface.co/mudler/all-MiniLM-L6-v2/resolve/main/ggml-model-q4_0.bin -O test-models/bert
 	wget -q https://cdn.openai.com/whisper/draft-20220913a/micro-machines.wav -O test-dir/audio.wav
@@ -417,6 +415,7 @@ protogen-python:
 	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
@@ -427,6 +426,7 @@ prepare-extra-conda-environments:
 	$(MAKE) -C backend/python/coqui
 	$(MAKE) -C backend/python/diffusers
 	$(MAKE) -C backend/python/vllm
+	$(MAKE) -C backend/python/mamba
 	$(MAKE) -C backend/python/sentencetransformers
 	$(MAKE) -C backend/python/transformers
 	$(MAKE) -C backend/python/transformers-musicgen
@@ -443,12 +443,18 @@ test-extra: prepare-test-extra
 	$(MAKE) -C backend/python/transformers test
 	$(MAKE) -C backend/python/diffusers test
 
+backend-assets:
+	mkdir -p backend-assets
+ifeq ($(BUILD_API_ONLY),true)
+	touch backend-assets/keep
+endif
+
 backend-assets/grpc:
 	mkdir -p backend-assets/grpc
 
 backend-assets/grpc/llama: backend-assets/grpc sources/go-llama/libbinding.a
-	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(shell pwd)/sources/go-llama
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-llama LIBRARY_PATH=$(shell pwd)/sources/go-llama \
+	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(CURDIR)/sources/go-llama
+	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/go-llama LIBRARY_PATH=$(CURDIR)/sources/go-llama \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/llama ./backend/go/llm/llama/
 # TODO: every binary should have its own folder instead, so can have different  implementations
 ifeq ($(BUILD_TYPE),metal)
@@ -467,17 +473,17 @@ ADDED_CMAKE_ARGS=-Dabsl_DIR=${INSTALLED_LIB_CMAKE}/absl \
 
 backend/cpp/llama/grpc-server:
 ifdef BUILD_GRPC_FOR_BACKEND_LLAMA
-	backend/cpp/grpc/script/build_grpc.sh ${INSTALLED_PACKAGES}
+	$(MAKE) -C backend/cpp/grpc build
 	export _PROTOBUF_PROTOC=${INSTALLED_PACKAGES}/bin/proto && \
 	export _GRPC_CPP_PLUGIN_EXECUTABLE=${INSTALLED_PACKAGES}/bin/grpc_cpp_plugin && \
-	export PATH=${PATH}:${INSTALLED_PACKAGES}/bin && \
+	export PATH="${INSTALLED_PACKAGES}/bin:${PATH}" && \
 	CMAKE_ARGS="${CMAKE_ARGS} ${ADDED_CMAKE_ARGS}" LLAMA_VERSION=$(CPPLLAMA_VERSION) $(MAKE) -C backend/cpp/llama grpc-server
 else
 	echo "BUILD_GRPC_FOR_BACKEND_LLAMA is not defined."
 	LLAMA_VERSION=$(CPPLLAMA_VERSION) $(MAKE) -C backend/cpp/llama grpc-server			
 endif
 ## BACKEND CPP LLAMA END
-		
+
 ##
 backend-assets/grpc/llama-cpp: backend-assets/grpc backend/cpp/llama/grpc-server
 	cp -rfv backend/cpp/llama/grpc-server backend-assets/grpc/llama-cpp
@@ -487,52 +493,20 @@ ifeq ($(BUILD_TYPE),metal)
 endif
 
 backend-assets/grpc/llama-ggml: backend-assets/grpc sources/go-llama-ggml/libbinding.a
-	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(shell pwd)/sources/go-llama-ggml
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-llama-ggml LIBRARY_PATH=$(shell pwd)/sources/go-llama-ggml \
+	$(GOCMD) mod edit -replace github.com/go-skynet/go-llama.cpp=$(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/gpt4all: backend-assets/grpc backend-assets/gpt4all sources/gpt4all/gpt4all-bindings/golang/libgpt4all.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/gpt4all/gpt4all-bindings/golang/ LIBRARY_PATH=$(shell pwd)/sources/gpt4all/gpt4all-bindings/golang/ \
+	CGO_LDFLAGS="$(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/dolly: backend-assets/grpc sources/go-ggml-transformers/libtransformers.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-ggml-transformers LIBRARY_PATH=$(shell pwd)/sources/go-ggml-transformers \
-	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/dolly ./backend/go/llm/dolly/
-
-backend-assets/grpc/gpt2: backend-assets/grpc sources/go-ggml-transformers/libtransformers.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-ggml-transformers LIBRARY_PATH=$(shell pwd)/sources/go-ggml-transformers \
-	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/gpt2 ./backend/go/llm/gpt2/
-
-backend-assets/grpc/gptj: backend-assets/grpc sources/go-ggml-transformers/libtransformers.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-ggml-transformers LIBRARY_PATH=$(shell pwd)/sources/go-ggml-transformers \
-	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/gptj ./backend/go/llm/gptj/
-
-backend-assets/grpc/gptneox: backend-assets/grpc sources/go-ggml-transformers/libtransformers.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-ggml-transformers LIBRARY_PATH=$(shell pwd)/sources/go-ggml-transformers \
-	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/gptneox ./backend/go/llm/gptneox/
-
-backend-assets/grpc/mpt: backend-assets/grpc sources/go-ggml-transformers/libtransformers.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-ggml-transformers LIBRARY_PATH=$(shell pwd)/sources/go-ggml-transformers \
-	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/mpt ./backend/go/llm/mpt/
-
-backend-assets/grpc/replit: backend-assets/grpc sources/go-ggml-transformers/libtransformers.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-ggml-transformers LIBRARY_PATH=$(shell pwd)/sources/go-ggml-transformers \
-	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/replit ./backend/go/llm/replit/
-
-backend-assets/grpc/falcon-ggml: backend-assets/grpc sources/go-ggml-transformers/libtransformers.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-ggml-transformers LIBRARY_PATH=$(shell pwd)/sources/go-ggml-transformers \
-	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/falcon-ggml ./backend/go/llm/falcon-ggml/
-
-backend-assets/grpc/starcoder: backend-assets/grpc sources/go-ggml-transformers/libtransformers.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-ggml-transformers LIBRARY_PATH=$(shell pwd)/sources/go-ggml-transformers \
-	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/starcoder ./backend/go/llm/starcoder/
-
 backend-assets/grpc/rwkv: backend-assets/grpc sources/go-rwkv/librwkv.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-rwkv LIBRARY_PATH=$(shell pwd)/sources/go-rwkv \
+	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/bert-embeddings: backend-assets/grpc sources/go-bert/libgobert.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-bert LIBRARY_PATH=$(shell pwd)/sources/go-bert \
+	CGO_LDFLAGS="$(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/langchain-huggingface: backend-assets/grpc
@@ -541,20 +515,39 @@ backend-assets/grpc/langchain-huggingface: backend-assets/grpc
 backend-assets/grpc/stablediffusion: backend-assets/grpc
 	if [ ! -f backend-assets/grpc/stablediffusion ]; then \
 		$(MAKE) sources/go-stable-diffusion/libstablediffusion.a; \
-		CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/go-stable-diffusion/ LIBRARY_PATH=$(shell pwd)/sources/go-stable-diffusion/ \
+		CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/go-stable-diffusion/ LIBRARY_PATH=$(CURDIR)/sources/go-stable-diffusion/ \
 		$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/stablediffusion ./backend/go/image/stablediffusion; \
 	fi
 
 backend-assets/grpc/tinydream: backend-assets/grpc sources/go-tiny-dream/libtinydream.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" LIBRARY_PATH=$(shell pwd)/go-tiny-dream \
+	CGO_LDFLAGS="$(CGO_LDFLAGS)" LIBRARY_PATH=$(CURDIR)/go-tiny-dream \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/tinydream ./backend/go/image/tinydream
 
 backend-assets/grpc/piper: backend-assets/grpc backend-assets/espeak-ng-data sources/go-piper/libpiper_binding.a
-	CGO_CXXFLAGS="$(PIPER_CGO_CXXFLAGS)" CGO_LDFLAGS="$(PIPER_CGO_LDFLAGS)" LIBRARY_PATH=$(shell pwd)/sources/go-piper \
+	CGO_CXXFLAGS="$(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/whisper: backend-assets/grpc sources/whisper.cpp/libwhisper.a
-	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(shell pwd)/sources/whisper.cpp LIBRARY_PATH=$(shell pwd)/sources/whisper.cpp \
+	CGO_LDFLAGS="$(CGO_LDFLAGS)" C_INCLUDE_PATH=$(CURDIR)/sources/whisper.cpp LIBRARY_PATH=$(CURDIR)/sources/whisper.cpp \
 	$(GOCMD) build -ldflags "$(LD_FLAGS)" -tags "$(GO_TAGS)" -o backend-assets/grpc/whisper ./backend/go/transcribe/
 
 grpcs: prepare $(GRPC_BACKENDS)
+
+DOCKER_IMAGE?=local-ai
+IMAGE_TYPE?=core
+BASE_IMAGE?=ubuntu:22.04
+
+docker:
+	docker build \
+		--build-arg BASE_IMAGE=$(BASE_IMAGE) \
+		--build-arg IMAGE_TYPE=$(IMAGE_TYPE) \
+		--build-arg GO_TAGS=$(GO_TAGS) \
+		--build-arg BUILD_TYPE=$(BUILD_TYPE) \
+		-t $(DOCKER_IMAGE) .
+
+docker-image-intel:
+	docker build \
+		--build-arg BASE_IMAGE=intel/oneapi-basekit:2024.0.1-devel-ubuntu22.04 \
+		--build-arg IMAGE_TYPE=$(IMAGE_TYPE) \
+		--build-arg GO_TAGS="none" \
+		--build-arg BUILD_TYPE=sycl_f16 -t $(DOCKER_IMAGE) .

README.md

@@ -43,6 +43,9 @@
 
 [Roadmap](https://github.com/mudler/LocalAI/issues?q=is%3Aissue+is%3Aopen+label%3Aroadmap)
 
+- Intel GPU support (sycl): https://github.com/mudler/LocalAI/issues/1653
+- Deprecation of old backends: https://github.com/mudler/LocalAI/issues/1651
+- 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
 - Inline templates: https://github.com/mudler/LocalAI/pull/1452
@@ -58,6 +61,12 @@ If you want to help and contribute, issues up for grabs: https://github.com/mudl
 
 ## 💻 [Getting started](https://localai.io/basics/getting_started/index.html)
 
+For a detailed step-by-step introduction, refer to the [Getting Started](https://localai.io/basics/getting_started/index.html) guide. For those in a hurry, here's a straightforward one-liner to launch a LocalAI instance with [phi-2](https://huggingface.co/microsoft/phi-2) using `docker`:
+
+```
+docker run -ti -p 8080:8080 localai/localai:v2.7.0-ffmpeg-core phi-2
+```
+
 ## 🚀 [Features](https://localai.io/features/)
 
 - 📖 [Text generation with GPTs](https://localai.io/features/text-generation/) (`llama.cpp`, `gpt4all.cpp`, ... [:book: and more](https://localai.io/model-compatibility/index.html#model-compatibility-table))
@@ -85,6 +94,10 @@ WebUIs:
 
 Model galleries
 - https://github.com/go-skynet/model-gallery
+  
+Auto Docker / Model setup
+- https://io.midori-ai.xyz/howtos/easy-localai-installer/
+- https://io.midori-ai.xyz/howtos/easy-model-installer/
 
 Other:
 - Helm chart https://github.com/go-skynet/helm-charts
@@ -98,11 +111,11 @@ Other:
 
 ### 🔗 Resources
 
-- 🆕 New! [LLM finetuning guide](https://localai.io/advanced/fine-tuning/)
+- 🆕 New! [LLM finetuning guide](https://localai.io/docs/advanced/fine-tuning/)
 - [How to build locally](https://localai.io/basics/build/index.html)
 - [How to install in Kubernetes](https://localai.io/basics/getting_started/index.html#run-localai-in-kubernetes)
-- [Projects integrating LocalAI](https://localai.io/integrations/)
-- [How tos section](https://localai.io/howtos/) (curated by our community)
+- [Projects integrating LocalAI](https://localai.io/docs/integrations/)
+- [How tos section](https://io.midori-ai.xyz/howtos/) (curated by our community)
 
 ## :book: 🎥 [Media, Blogs, Social](https://localai.io/basics/news/#media-blogs-social)
 
@@ -165,7 +178,6 @@ LocalAI couldn't have been built without the help of great software already avai
 - https://github.com/ggerganov/whisper.cpp
 - https://github.com/saharNooby/rwkv.cpp
 - https://github.com/rhasspy/piper
-- https://github.com/cmp-nct/ggllm.cpp
 
 ## 🤗 Contributors
 

api/api.go

@@ -37,7 +37,7 @@ func Startup(opts ...options.AppOption) (*options.Option, *config.ConfigLoader,
 	log.Info().Msgf("Starting LocalAI using %d threads, with models path: %s", options.Threads, options.Loader.ModelPath)
 	log.Info().Msgf("LocalAI version: %s", internal.PrintableVersion())
 
-	startup.PreloadModelsConfigurations(options.Loader.ModelPath, options.ModelsURL...)
+	startup.PreloadModelsConfigurations(options.ModelLibraryURL, options.Loader.ModelPath, options.ModelsURL...)
 
 	cl := config.NewConfigLoader()
 	if err := cl.LoadConfigs(options.Loader.ModelPath); err != nil {
@@ -216,6 +216,11 @@ func App(opts ...options.AppOption) (*fiber.App, error) {
 		}{Version: internal.PrintableVersion()})
 	})
 
+	// Make sure directories exists
+	os.MkdirAll(options.ImageDir, 0755)
+	os.MkdirAll(options.AudioDir, 0755)
+	os.MkdirAll(options.Loader.ModelPath, 0755)
+
 	modelGalleryService := localai.CreateModelGalleryService(options.Galleries, options.Loader.ModelPath, galleryService)
 	app.Post("/models/apply", auth, modelGalleryService.ApplyModelGalleryEndpoint())
 	app.Get("/models/available", auth, modelGalleryService.ListModelFromGalleryEndpoint())

api/api_test.go

@@ -29,6 +29,15 @@ import (
 	"github.com/sashabaranov/go-openai/jsonschema"
 )
 
+const testPrompt = `### System:
+You are an AI assistant that follows instruction extremely well. Help as much as you can.
+
+### User:
+
+Can you help rephrasing sentences?
+
+### Response:`
+
 type modelApplyRequest struct {
 	ID        string                 `json:"id"`
 	URL       string                 `json:"url"`
@@ -629,28 +638,28 @@ var _ = Describe("API test", func() {
 			Expect(len(models.Models)).To(Equal(6)) // If "config.yaml" should be included, this should be 8?
 		})
 		It("can generate completions", func() {
-			resp, err := client.CreateCompletion(context.TODO(), openai.CompletionRequest{Model: "testmodel", Prompt: "abcdedfghikl"})
+			resp, err := client.CreateCompletion(context.TODO(), openai.CompletionRequest{Model: "testmodel", Prompt: testPrompt})
 			Expect(err).ToNot(HaveOccurred())
 			Expect(len(resp.Choices)).To(Equal(1))
 			Expect(resp.Choices[0].Text).ToNot(BeEmpty())
 		})
 
 		It("can generate chat completions ", func() {
-			resp, err := client.CreateChatCompletion(context.TODO(), openai.ChatCompletionRequest{Model: "testmodel", Messages: []openai.ChatCompletionMessage{openai.ChatCompletionMessage{Role: "user", Content: "abcdedfghikl"}}})
+			resp, err := client.CreateChatCompletion(context.TODO(), openai.ChatCompletionRequest{Model: "testmodel", Messages: []openai.ChatCompletionMessage{openai.ChatCompletionMessage{Role: "user", Content: testPrompt}}})
 			Expect(err).ToNot(HaveOccurred())
 			Expect(len(resp.Choices)).To(Equal(1))
 			Expect(resp.Choices[0].Message.Content).ToNot(BeEmpty())
 		})
 
 		It("can generate completions from model configs", func() {
-			resp, err := client.CreateCompletion(context.TODO(), openai.CompletionRequest{Model: "gpt4all", Prompt: "abcdedfghikl"})
+			resp, err := client.CreateCompletion(context.TODO(), openai.CompletionRequest{Model: "gpt4all", Prompt: testPrompt})
 			Expect(err).ToNot(HaveOccurred())
 			Expect(len(resp.Choices)).To(Equal(1))
 			Expect(resp.Choices[0].Text).ToNot(BeEmpty())
 		})
 
 		It("can generate chat completions from model configs", func() {
-			resp, err := client.CreateChatCompletion(context.TODO(), openai.ChatCompletionRequest{Model: "gpt4all-2", Messages: []openai.ChatCompletionMessage{openai.ChatCompletionMessage{Role: "user", Content: "abcdedfghikl"}}})
+			resp, err := client.CreateChatCompletion(context.TODO(), openai.ChatCompletionRequest{Model: "gpt4all-2", Messages: []openai.ChatCompletionMessage{openai.ChatCompletionMessage{Role: "user", Content: testPrompt}}})
 			Expect(err).ToNot(HaveOccurred())
 			Expect(len(resp.Choices)).To(Equal(1))
 			Expect(resp.Choices[0].Message.Content).ToNot(BeEmpty())
@@ -658,7 +667,7 @@ var _ = Describe("API test", func() {
 
 		It("returns errors", func() {
 			backends := len(model.AutoLoadBackends) + 1 // +1 for huggingface
-			_, err := client.CreateCompletion(context.TODO(), openai.CompletionRequest{Model: "foomodel", Prompt: "abcdedfghikl"})
+			_, err := client.CreateCompletion(context.TODO(), openai.CompletionRequest{Model: "foomodel", Prompt: testPrompt})
 			Expect(err).To(HaveOccurred())
 			Expect(err.Error()).To(ContainSubstring(fmt.Sprintf("error, status code: 500, message: could not load model - all backends returned error: %d errors occurred:", backends)))
 		})
@@ -834,13 +843,13 @@ var _ = Describe("API test", func() {
 			app.Shutdown()
 		})
 		It("can generate chat completions from config file (list1)", func() {
-			resp, err := client.CreateChatCompletion(context.TODO(), openai.ChatCompletionRequest{Model: "list1", Messages: []openai.ChatCompletionMessage{{Role: "user", Content: "abcdedfghikl"}}})
+			resp, err := client.CreateChatCompletion(context.TODO(), openai.ChatCompletionRequest{Model: "list1", Messages: []openai.ChatCompletionMessage{{Role: "user", Content: testPrompt}}})
 			Expect(err).ToNot(HaveOccurred())
 			Expect(len(resp.Choices)).To(Equal(1))
 			Expect(resp.Choices[0].Message.Content).ToNot(BeEmpty())
 		})
 		It("can generate chat completions from config file (list2)", func() {
-			resp, err := client.CreateChatCompletion(context.TODO(), openai.ChatCompletionRequest{Model: "list2", Messages: []openai.ChatCompletionMessage{{Role: "user", Content: "abcdedfghikl"}}})
+			resp, err := client.CreateChatCompletion(context.TODO(), openai.ChatCompletionRequest{Model: "list2", Messages: []openai.ChatCompletionMessage{{Role: "user", Content: testPrompt}}})
 			Expect(err).ToNot(HaveOccurred())
 			Expect(len(resp.Choices)).To(Equal(1))
 			Expect(resp.Choices[0].Message.Content).ToNot(BeEmpty())

api/backend/embeddings.go

@@ -41,7 +41,7 @@ func ModelEmbedding(s string, tokens []int, loader *model.ModelLoader, c config.
 
 	var fn func() ([]float32, error)
 	switch model := inferenceModel.(type) {
-	case *grpc.Client:
+	case grpc.Backend:
 		fn = func() ([]float32, error) {
 			predictOptions := gRPCPredictOpts(c, loader.ModelPath)
 			if len(tokens) > 0 {

api/backend/llm.go

@@ -31,7 +31,7 @@ func ModelInference(ctx context.Context, s string, images []string, loader *mode
 
 	grpcOpts := gRPCModelOpts(c)
 
-	var inferenceModel *grpc.Client
+	var inferenceModel grpc.Backend
 	var err error
 
 	opts := modelOpts(c, o, []model.Option{

api/backend/options.go

@@ -63,6 +63,8 @@ func gRPCModelOpts(c config.Config) *pb.ModelOptions {
 		F16Memory:      c.F16,
 		MLock:          c.MMlock,
 		RopeFreqBase:   c.RopeFreqBase,
+		RopeScaling:    c.RopeScaling,
+		Type:           c.ModelType,
 		RopeFreqScale:  c.RopeFreqScale,
 		NUMA:           c.NUMA,
 		Embeddings:     c.Embeddings,

api/config/config.go

@@ -55,6 +55,9 @@ type Config struct {
 	CUDA bool `yaml:"cuda"`
 
 	DownloadFiles []File `yaml:"download_files"`
+
+	Description string `yaml:"description"`
+	Usage       string `yaml:"usage"`
 }
 
 type File struct {
@@ -125,7 +128,9 @@ type LLMConfig struct {
 	Quantization string  `yaml:"quantization"`
 	MMProj       string  `yaml:"mmproj"`
 
-	RopeScaling    string  `yaml:"rope_scaling"`
+	RopeScaling string `yaml:"rope_scaling"`
+	ModelType   string `yaml:"type"`
+
 	YarnExtFactor  float32 `yaml:"yarn_ext_factor"`
 	YarnAttnFactor float32 `yaml:"yarn_attn_factor"`
 	YarnBetaFast   float32 `yaml:"yarn_beta_fast"`
@@ -326,6 +331,15 @@ func (cm *ConfigLoader) Preload(modelPath string) error {
 			c.PredictionOptions.Model = md5Name
 			cm.configs[i] = *c
 		}
+		if cm.configs[i].Name != "" {
+			log.Info().Msgf("Model name: %s", cm.configs[i].Name)
+		}
+		if cm.configs[i].Description != "" {
+			log.Info().Msgf("Model description: %s", cm.configs[i].Description)
+		}
+		if cm.configs[i].Usage != "" {
+			log.Info().Msgf("Model usage: \n%s", cm.configs[i].Usage)
+		}
 	}
 	return nil
 }

api/options/options.go

@@ -28,6 +28,8 @@ type Option struct {
 	ApiKeys                             []string
 	Metrics                             *metrics.Metrics
 
+	ModelLibraryURL string
+
 	Galleries []gallery.Gallery
 
 	BackendAssets     embed.FS
@@ -78,6 +80,12 @@ func WithCors(b bool) AppOption {
 	}
 }
 
+func WithModelLibraryURL(url string) AppOption {
+	return func(o *Option) {
+		o.ModelLibraryURL = url
+	}
+}
+
 var EnableWatchDog = func(o *Option) {
 	o.WatchDog = true
 }

backend/backend.proto

@@ -112,7 +112,6 @@ message ModelOptions {
   int32 CLIPSkip = 33;
   string ControlNet = 48;
 
-  // RWKV
   string Tokenizer = 34;
 
   // LLM (llama.cpp)
@@ -135,6 +134,8 @@ message ModelOptions {
   float YarnAttnFactor = 45;
   float YarnBetaFast = 46;
   float YarnBetaSlow = 47;
+
+  string Type = 49;
 }
 
 message Result {

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

@@ -0,0 +1,66 @@
+# Basic platform detection
+HOST_SYSTEM = $(shell uname | cut -f 1 -d_)
+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
+GRPC_BUILD=grpc_build
+
+export CMAKE_ARGS?=
+CMAKE_ARGS+=-DCMAKE_BUILD_TYPE=Release
+CMAKE_ARGS+=-DgRPC_INSTALL=ON
+CMAKE_ARGS+=-DEXECUTABLE_OUTPUT_PATH=../$(INSTALLED_PACKAGES)/grpc/bin
+CMAKE_ARGS+=-DLIBRARY_OUTPUT_PATH=../$(INSTALLED_PACKAGES)/grpc/lib
+CMAKE_ARGS+=-DgRPC_BUILD_TESTS=OFF
+CMAKE_ARGS+=-DgRPC_BUILD_CSHARP_EXT=OFF
+CMAKE_ARGS+=-DgRPC_BUILD_GRPC_CPP_PLUGIN=ON
+CMAKE_ARGS+=-DgRPC_BUILD_GRPC_CSHARP_PLUGIN=OFF
+CMAKE_ARGS+=-DgRPC_BUILD_GRPC_NODE_PLUGIN=OFF
+CMAKE_ARGS+=-DgRPC_BUILD_GRPC_OBJECTIVE_C_PLUGIN=OFF
+CMAKE_ARGS+=-DgRPC_BUILD_GRPC_PHP_PLUGIN=OFF
+CMAKE_ARGS+=-DgRPC_BUILD_GRPC_PYTHON_PLUGIN=ON
+CMAKE_ARGS+=-DgRPC_BUILD_GRPC_RUBY_PLUGIN=OFF
+CMAKE_ARGS+=-Dprotobuf_WITH_ZLIB=ON
+CMAKE_ARGS+=-DRE2_BUILD_TESTING=OFF
+CMAKE_ARGS+=-DCMAKE_INSTALL_PREFIX=../$(INSTALLED_PACKAGES)
+
+# windows need to set OPENSSL_NO_ASM. Results in slower crypto performance but doesn't build otherwise.
+# May be resolvable, but for now its set. More info: https://stackoverflow.com/a/75240504/480673
+ifeq ($(SYSTEM),MSYS)
+CMAKE_ARGS+=-DOPENSSL_NO_ASM=ON
+endif
+ifeq ($(SYSTEM),MINGW64)
+CMAKE_ARGS+=-DOPENSSL_NO_ASM=ON
+endif
+ifeq ($(SYSTEM),MINGW32)
+CMAKE_ARGS+=-DOPENSSL_NO_ASM=ON
+endif
+ifeq ($(SYSTEM),CYGWIN)
+CMAKE_ARGS+=-DOPENSSL_NO_ASM=ON
+endif
+
+$(INSTALLED_PACKAGES): grpc_build
+
+$(GRPC_REPO):
+	git clone --depth $(GIT_CLONE_DEPTH) -b $(TAG_LIB_GRPC) $(GIT_REPO_LIB_GRPC) $(GRPC_REPO)/grpc
+	cd $(GRPC_REPO)/grpc && git submodule update --init --recursive --depth $(GIT_CLONE_DEPTH)
+
+$(GRPC_BUILD): $(GRPC_REPO)
+	mkdir -p $(GRPC_BUILD)
+	cd $(GRPC_BUILD) && cmake $(CMAKE_ARGS) ../$(GRPC_REPO)/grpc && cmake --build . -- -j ${NUM_BUILD_THREADS} && cmake --build . --target install -- -j ${NUM_BUILD_THREADS}
+
+build: $(INSTALLED_PACKAGES)
+
+rebuild:
+	rm -rf grpc_build
+	$(MAKE) grpc_build
+
+clean:
+	rm -rf grpc_build
+	rm -rf grpc_repo
+	rm -rf installed_packages

backend/cpp/grpc/script/build_grpc.sh

@@ -1,81 +0,0 @@
-#!/bin/bash
-
-# Builds locally from sources the packages needed by the llama cpp backend.
-
-# Makes sure a few base packages exist.
-# sudo apt-get --no-upgrade -y install g++ gcc binutils cmake git build-essential autoconf libtool pkg-config 
-
-SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )"
-echo "Script directory: $SCRIPT_DIR"
-
-CPP_INSTALLED_PACKAGES_DIR=$1
-if [ -z ${CPP_INSTALLED_PACKAGES_DIR} ]; then 
-    echo "CPP_INSTALLED_PACKAGES_DIR env variable not set. Don't know where to install: failed."; 
-    echo
-    exit -1
-fi
-
-if [ -d "${CPP_INSTALLED_PACKAGES_DIR}" ]; then
-  echo "gRPC installation directory already exists. Nothing to do."
-  exit 0
-fi
-
-# The depth when cloning a git repo. 1 speeds up the clone when the repo history is not needed.
-GIT_CLONE_DEPTH=1
-
-NUM_BUILD_THREADS=$(nproc --ignore=1)
-
-# Google gRPC --------------------------------------------------------------------------------------
-TAG_LIB_GRPC="v1.59.0"
-GIT_REPO_LIB_GRPC="https://github.com/grpc/grpc.git"
-GRPC_REPO_DIR="${SCRIPT_DIR}/../grpc_repo"
-GRPC_BUILD_DIR="${SCRIPT_DIR}/../grpc_build"
-SRC_DIR_LIB_GRPC="${GRPC_REPO_DIR}/grpc"
-
-echo "SRC_DIR_LIB_GRPC: ${SRC_DIR_LIB_GRPC}"
-echo "GRPC_REPO_DIR: ${GRPC_REPO_DIR}"
-echo "GRPC_BUILD_DIR: ${GRPC_BUILD_DIR}"
-
-mkdir -pv ${GRPC_REPO_DIR}
-
-rm   -rf ${GRPC_BUILD_DIR}
-mkdir -pv ${GRPC_BUILD_DIR}
-
-mkdir -pv ${CPP_INSTALLED_PACKAGES_DIR}
-	
-if [ -d "${SRC_DIR_LIB_GRPC}" ]; then
-  echo "gRPC source already exists locally. Not cloned again."
-else  
-  ( cd ${GRPC_REPO_DIR} && \
-    git clone --depth ${GIT_CLONE_DEPTH} -b ${TAG_LIB_GRPC} ${GIT_REPO_LIB_GRPC} && \
-    cd ${SRC_DIR_LIB_GRPC} && \
-    git submodule update --init --recursive --depth ${GIT_CLONE_DEPTH} 
-  )    
-fi
-
-( cd ${GRPC_BUILD_DIR} && \
-  cmake -G "Unix Makefiles" \
-     -DCMAKE_BUILD_TYPE=Release \
-     -DgRPC_INSTALL=ON \
-     -DEXECUTABLE_OUTPUT_PATH=${CPP_INSTALLED_PACKAGES_DIR}/grpc/bin \
-     -DLIBRARY_OUTPUT_PATH=${CPP_INSTALLED_PACKAGES_DIR}/grpc/lib \
-     -DgRPC_BUILD_TESTS=OFF \
-     -DgRPC_BUILD_CSHARP_EXT=OFF \
-     -DgRPC_BUILD_GRPC_CPP_PLUGIN=ON \
-     -DgRPC_BUILD_GRPC_CSHARP_PLUGIN=OFF \
-     -DgRPC_BUILD_GRPC_NODE_PLUGIN=OFF \
-     -DgRPC_BUILD_GRPC_OBJECTIVE_C_PLUGIN=OFF \
-     -DgRPC_BUILD_GRPC_PHP_PLUGIN=OFF \
-     -DgRPC_BUILD_GRPC_PYTHON_PLUGIN=ON \
-     -DgRPC_BUILD_GRPC_RUBY_PLUGIN=OFF \
-     -Dprotobuf_WITH_ZLIB=ON \
-     -DRE2_BUILD_TESTING=OFF \
-     -DCMAKE_INSTALL_PREFIX=${CPP_INSTALLED_PACKAGES_DIR}/ \
-     ${SRC_DIR_LIB_GRPC}  && \
-  cmake --build .  -- -j ${NUM_BUILD_THREADS} && \
-  cmake --build .  --target install -- -j ${NUM_BUILD_THREADS} 
-)
-
-rm -rf ${GRPC_BUILD_DIR}
-rm -rf ${GRPC_REPO_DIR}
-

backend/cpp/llama/CMakeLists.txt

@@ -70,7 +70,7 @@ add_library(hw_grpc_proto
   ${hw_proto_srcs}
   ${hw_proto_hdrs} )
 
-add_executable(${TARGET} grpc-server.cpp json.hpp )
+add_executable(${TARGET} grpc-server.cpp utils.hpp json.hpp)
 target_link_libraries(${TARGET} PRIVATE common llama myclip ${CMAKE_THREAD_LIBS_INIT} absl::flags hw_grpc_proto
   absl::flags_parse
   gRPC::${_REFLECTION}

backend/cpp/llama/Makefile

@@ -3,6 +3,7 @@ LLAMA_VERSION?=
 
 CMAKE_ARGS?=
 BUILD_TYPE?=
+ONEAPI_VARS?=/opt/intel/oneapi/setvars.sh
 
 # If build type is cublas, then we set -DLLAMA_CUBLAS=ON to CMAKE_ARGS automatically
 ifeq ($(BUILD_TYPE),cublas)
@@ -19,6 +20,14 @@ else ifeq ($(BUILD_TYPE),hipblas)
 	CMAKE_ARGS+=-DLLAMA_HIPBLAS=ON
 endif
 
+ifeq ($(BUILD_TYPE),sycl_f16)
+	CMAKE_ARGS+=-DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
+endif
+
+ifeq ($(BUILD_TYPE),sycl_f32)
+	CMAKE_ARGS+=-DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
+endif
+
 llama.cpp:
 	git clone --recurse-submodules https://github.com/ggerganov/llama.cpp llama.cpp
 	if [ -z "$(LLAMA_VERSION)" ]; then \
@@ -31,6 +40,7 @@ llama.cpp/examples/grpc-server:
 	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.
@@ -49,5 +59,10 @@ clean:
 	rm -rf grpc-server
 
 grpc-server: llama.cpp llama.cpp/examples/grpc-server
+ifneq (,$(findstring sycl,$(BUILD_TYPE)))
+	bash -c "source $(ONEAPI_VARS); \
+	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) && cmake --build . --config Release
+endif
 	cp llama.cpp/build/bin/grpc-server .

backend/cpp/llama/utils.hpp

@@ -0,0 +1,510 @@
+// https://github.com/ggerganov/llama.cpp/blob/master/examples/server/utils.hpp
+
+#pragma once
+
+#include <string>
+#include <vector>
+#include <set>
+#include <mutex>
+#include <condition_variable>
+#include <unordered_map>
+
+#include "json.hpp"
+
+#include "../llava/clip.h"
+
+using json = nlohmann::json;
+
+extern bool server_verbose;
+
+#ifndef SERVER_VERBOSE
+#define SERVER_VERBOSE 1
+#endif
+
+#if SERVER_VERBOSE != 1
+#define LOG_VERBOSE(MSG, ...)
+#else
+#define LOG_VERBOSE(MSG, ...)                                            \
+    do                                                                   \
+    {                                                                    \
+        if (server_verbose)                                              \
+        {                                                                \
+            server_log("VERBOSE", __func__, __LINE__, MSG, __VA_ARGS__); \
+        }                                                                \
+    } while (0)
+#endif
+
+#define LOG_ERROR(  MSG, ...) server_log("ERROR",   __func__, __LINE__, MSG, __VA_ARGS__)
+#define LOG_WARNING(MSG, ...) server_log("WARNING", __func__, __LINE__, MSG, __VA_ARGS__)
+#define LOG_INFO(   MSG, ...) server_log("INFO",    __func__, __LINE__, MSG, __VA_ARGS__)
+
+//
+// parallel
+//
+
+enum server_state {
+    SERVER_STATE_LOADING_MODEL,  // Server is starting up, model not fully loaded yet
+    SERVER_STATE_READY,          // Server is ready and model is loaded
+    SERVER_STATE_ERROR           // An error occurred, load_model failed
+};
+
+enum task_type {
+    TASK_TYPE_COMPLETION,
+    TASK_TYPE_CANCEL,
+    TASK_TYPE_NEXT_RESPONSE
+};
+
+struct task_server {
+    int id = -1; // to be filled by llama_server_queue
+    int target_id;
+    task_type type;
+    json data;
+    bool infill_mode = false;
+    bool embedding_mode = false;
+    int multitask_id = -1;
+};
+
+struct task_result {
+    int id;
+    int multitask_id = -1;
+    bool stop;
+    bool error;
+    json result_json;
+};
+
+struct task_multi {
+    int id;
+    std::set<int> subtasks_remaining{};
+    std::vector<task_result> results{};
+};
+
+// TODO: can become bool if we can't find use of more states
+enum slot_state
+{
+    IDLE,
+    PROCESSING,
+};
+
+enum slot_command
+{
+    NONE,
+    LOAD_PROMPT,
+    RELEASE,
+};
+
+struct slot_params
+{
+    bool stream       = true;
+    bool cache_prompt = false; // remember the prompt to avoid reprocessing all prompt
+
+    uint32_t seed      = -1; // RNG seed
+    int32_t  n_keep    =  0; // number of tokens to keep from initial prompt
+    int32_t  n_predict = -1; // new tokens to predict
+
+    std::vector<std::string> antiprompt;
+
+    json input_prefix;
+    json input_suffix;
+};
+
+struct slot_image
+{
+    int32_t id;
+
+    bool request_encode_image = false;
+    float * image_embedding = nullptr;
+    int32_t image_tokens = 0;
+
+    clip_image_u8 * img_data;
+
+    std::string prefix_prompt; // before of this image
+};
+
+// completion token output with probabilities
+struct completion_token_output
+{
+    struct token_prob
+    {
+        llama_token tok;
+        float prob;
+    };
+
+    std::vector<token_prob> probs;
+    llama_token tok;
+    std::string text_to_send;
+};
+
+static inline void server_log(const char *level, const char *function, int line,
+                       const char *message, const nlohmann::ordered_json &extra)
+{
+    nlohmann::ordered_json log
+    {
+        {"timestamp", time(nullptr)},
+        {"level",     level},
+        {"function",  function},
+        {"line",      line},
+        {"message",   message},
+    };
+
+    if (!extra.empty())
+    {
+        log.merge_patch(extra);
+    }
+
+    const std::string str = log.dump(-1, ' ', false, json::error_handler_t::replace);
+    printf("%.*s\n", (int)str.size(), str.data());
+    fflush(stdout);
+}
+
+//
+// server utils
+//
+
+template <typename T>
+static T json_value(const json &body, const std::string &key, const T &default_value)
+{
+    // Fallback null to default value
+    return body.contains(key) && !body.at(key).is_null()
+        ? body.value(key, default_value)
+        : default_value;
+}
+
+inline std::string format_chatml(std::vector<json> messages)
+{
+    std::ostringstream chatml_msgs;
+
+    for (auto it = messages.begin(); it != messages.end(); ++it) {
+        chatml_msgs << "<|im_start|>"
+                    << json_value(*it, "role",    std::string("user")) << '\n';
+        chatml_msgs << json_value(*it, "content", std::string(""))
+                    << "<|im_end|>\n";
+    }
+
+    chatml_msgs << "<|im_start|>assistant" << '\n';
+
+    return chatml_msgs.str();
+}
+
+//
+// work queue utils
+//
+
+struct llama_server_queue {
+    int id = 0;
+    std::mutex mutex_tasks;
+    // queues
+    std::vector<task_server> queue_tasks;
+    std::vector<task_server> queue_tasks_deferred;
+    std::vector<task_multi> queue_multitasks;
+    std::condition_variable condition_tasks;
+    // callback functions
+    std::function<void(task_server&)> callback_new_task;
+    std::function<void(task_multi&)> callback_finish_multitask;
+    std::function<void(void)> callback_all_task_finished;
+
+    // Add a new task to the end of the queue
+    int post(task_server task) {
+        std::unique_lock<std::mutex> lock(mutex_tasks);
+        if (task.id == -1) {
+            task.id = id++;
+        }
+        queue_tasks.push_back(std::move(task));
+        condition_tasks.notify_one();
+        return task.id;
+    }
+
+    // Add a new task, but defer until one slot is available
+    void defer(task_server task) {
+        std::unique_lock<std::mutex> lock(mutex_tasks);
+        queue_tasks_deferred.push_back(std::move(task));
+    }
+
+    // Get the next id for creating anew task
+    int get_new_id() {
+        std::unique_lock<std::mutex> lock(mutex_tasks);
+        return id++;
+    }
+
+    // Register function to process a new task
+    void on_new_task(std::function<void(task_server&)> callback) {
+        callback_new_task = callback;
+    }
+
+    // Register function to process a multitask
+    void on_finish_multitask(std::function<void(task_multi&)> callback) {
+        callback_finish_multitask = callback;
+    }
+
+    // Register the function to be called when the batch of tasks is finished
+    void on_all_tasks_finished(std::function<void(void)> callback) {
+        callback_all_task_finished = callback;
+    }
+
+    // Call when the state of one slot is changed
+    void notify_slot_changed() {
+        // move deferred tasks back to main loop
+        std::unique_lock<std::mutex> lock(mutex_tasks);
+        for (auto & task : queue_tasks_deferred) {
+            queue_tasks.push_back(std::move(task));
+        }
+        queue_tasks_deferred.clear();
+    }
+
+    // Start the main loop. This call is blocking
+    [[noreturn]]
+    void start_loop() {
+        while (true) {
+            // new task arrived
+            LOG_VERBOSE("have new task", {});
+            {
+                while (true)
+                {
+                    std::unique_lock<std::mutex> lock(mutex_tasks);
+                    if (queue_tasks.empty()) {
+                        lock.unlock();
+                        break;
+                    }
+                    task_server task = queue_tasks.front();
+                    queue_tasks.erase(queue_tasks.begin());
+                    lock.unlock();
+                    LOG_VERBOSE("callback_new_task", {});
+                    callback_new_task(task);
+                }
+                LOG_VERBOSE("callback_all_task_finished", {});
+                // process and update all the multitasks
+                auto queue_iterator = queue_multitasks.begin();
+                while (queue_iterator != queue_multitasks.end())
+                {
+                    if (queue_iterator->subtasks_remaining.empty())
+                    {
+                        // all subtasks done == multitask is done
+                        task_multi current_multitask = *queue_iterator;
+                        callback_finish_multitask(current_multitask);
+                        // remove this multitask
+                        queue_iterator = queue_multitasks.erase(queue_iterator);
+                    }
+                    else
+                    {
+                        ++queue_iterator;
+                    }
+                }
+                // all tasks in the current loop is finished
+                callback_all_task_finished();
+            }
+            LOG_VERBOSE("wait for new task", {});
+            // wait for new task
+            {
+                std::unique_lock<std::mutex> lock(mutex_tasks);
+                if (queue_tasks.empty()) {
+                    condition_tasks.wait(lock, [&]{
+                        return !queue_tasks.empty();
+                    });
+                }
+            }
+        }
+    }
+
+    //
+    // functions to manage multitasks
+    //
+
+    // add a multitask by specifying the id of all subtask (subtask is a task_server)
+    void add_multitask(int multitask_id, std::vector<int>& sub_ids)
+    {
+        std::lock_guard<std::mutex> lock(mutex_tasks);
+        task_multi multi;
+        multi.id = multitask_id;
+        std::copy(sub_ids.begin(), sub_ids.end(), std::inserter(multi.subtasks_remaining, multi.subtasks_remaining.end()));
+        queue_multitasks.push_back(multi);
+    }
+
+    // updatethe remaining subtasks, while appending results to multitask
+    void update_multitask(int multitask_id, int subtask_id, task_result& result)
+    {
+        std::lock_guard<std::mutex> lock(mutex_tasks);
+        for (auto& multitask : queue_multitasks)
+        {
+            if (multitask.id == multitask_id)
+            {
+                multitask.subtasks_remaining.erase(subtask_id);
+                multitask.results.push_back(result);
+            }
+        }
+    }
+};
+
+struct llama_server_response {
+    typedef std::function<void(int, int, task_result&)> callback_multitask_t;
+    callback_multitask_t callback_update_multitask;
+    // for keeping track of all tasks waiting for the result
+    std::set<int> waiting_task_ids;
+    // the main result queue
+    std::vector<task_result> queue_results;
+    std::mutex mutex_results;
+    std::condition_variable condition_results;
+
+    void add_waiting_task_id(int task_id) {
+        std::unique_lock<std::mutex> lock(mutex_results);
+        waiting_task_ids.insert(task_id);
+    }
+
+    void remove_waiting_task_id(int task_id) {
+        std::unique_lock<std::mutex> lock(mutex_results);
+        waiting_task_ids.erase(task_id);
+    }
+
+    // This function blocks the thread until there is a response for this task_id
+    task_result recv(int task_id) {
+        while (true)
+        {
+            std::unique_lock<std::mutex> lock(mutex_results);
+            condition_results.wait(lock, [&]{
+                return !queue_results.empty();
+            });
+            LOG_VERBOSE("condition_results unblock", {});
+
+            for (int i = 0; i < (int) queue_results.size(); i++)
+            {
+                if (queue_results[i].id == task_id)
+                {
+                    assert(queue_results[i].multitask_id == -1);
+                    task_result res = queue_results[i];
+                    queue_results.erase(queue_results.begin() + i);
+                    return res;
+                }
+            }
+        }
+
+        // should never reach here
+    }
+
+    // Register the function to update multitask
+    void on_multitask_update(callback_multitask_t callback) {
+        callback_update_multitask = callback;
+    }
+
+    // Send a new result to a waiting task_id
+    void send(task_result result) {
+        std::unique_lock<std::mutex> lock(mutex_results);
+        LOG_VERBOSE("send new result", {});
+        for (auto& task_id : waiting_task_ids) {
+            // LOG_TEE("waiting task id %i \n", task_id);
+            // for now, tasks that have associated parent multitasks just get erased once multitask picks up the result
+            if (result.multitask_id == task_id)
+            {
+                LOG_VERBOSE("callback_update_multitask", {});
+                callback_update_multitask(task_id, result.id, result);
+                continue;
+            }
+
+            if (result.id == task_id)
+            {
+                LOG_VERBOSE("queue_results.push_back", {});
+                queue_results.push_back(result);
+                condition_results.notify_one();
+                return;
+            }
+        }
+    }
+};
+
+//
+// base64 utils (TODO: move to common in the future)
+//
+
+static const std::string base64_chars =
+             "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+             "abcdefghijklmnopqrstuvwxyz"
+             "0123456789+/";
+
+static inline bool is_base64(uint8_t c)
+{
+    return (isalnum(c) || (c == '+') || (c == '/'));
+}
+
+static inline std::vector<uint8_t> base64_decode(const std::string & encoded_string)
+{
+    int i = 0;
+    int j = 0;
+    int in_ = 0;
+
+    int in_len = encoded_string.size();
+
+    uint8_t char_array_4[4];
+    uint8_t char_array_3[3];
+
+    std::vector<uint8_t> ret;
+
+    while (in_len-- && (encoded_string[in_] != '=') && is_base64(encoded_string[in_]))
+    {
+        char_array_4[i++] = encoded_string[in_]; in_++;
+        if (i == 4)
+        {
+            for (i = 0; i <4; i++)
+            {
+                char_array_4[i] = base64_chars.find(char_array_4[i]);
+            }
+
+            char_array_3[0] = ((char_array_4[0]      ) << 2) + ((char_array_4[1] & 0x30) >> 4);
+            char_array_3[1] = ((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2);
+            char_array_3[2] = ((char_array_4[2] & 0x3) << 6) +   char_array_4[3];
+
+            for (i = 0; (i < 3); i++)
+            {
+                ret.push_back(char_array_3[i]);
+            }
+            i = 0;
+        }
+    }
+
+    if (i)
+    {
+        for (j = i; j <4; j++)
+        {
+            char_array_4[j] = 0;
+        }
+
+        for (j = 0; j <4; j++)
+        {
+            char_array_4[j] = base64_chars.find(char_array_4[j]);
+        }
+
+        char_array_3[0] = ((char_array_4[0]      ) << 2) + ((char_array_4[1] & 0x30) >> 4);
+        char_array_3[1] = ((char_array_4[1] & 0xf) << 4) + ((char_array_4[2] & 0x3c) >> 2);
+        char_array_3[2] = ((char_array_4[2] & 0x3) << 6) +   char_array_4[3];
+
+        for (j = 0; (j < i - 1); j++)
+        {
+            ret.push_back(char_array_3[j]);
+        }
+    }
+
+    return ret;
+}
+
+//
+// random string / id
+//
+
+static std::string random_string()
+{
+    static const std::string str("0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz");
+
+    std::random_device rd;
+    std::mt19937 generator(rd());
+
+    std::string result(32, ' ');
+
+    for (int i = 0; i < 32; ++i) {
+        result[i] = str[generator() % str.size()];
+    }
+
+    return result;
+}
+
+static std::string gen_chatcmplid()
+{
+    std::stringstream chatcmplid;
+    chatcmplid << "chatcmpl-" << random_string();
+    return chatcmplid.str();
+}

backend/go/llm/falcon-ggml/main.go

@@ -1,23 +0,0 @@
-package main
-
-// Note: this is started internally by LocalAI and a server is allocated for each model
-
-import (
-	"flag"
-
-	transformers "github.com/go-skynet/LocalAI/backend/go/llm/transformers"
-
-	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
-)
-
-var (
-	addr = flag.String("addr", "localhost:50051", "the address to connect to")
-)
-
-func main() {
-	flag.Parse()
-
-	if err := grpc.StartServer(*addr, &transformers.Falcon{}); err != nil {
-		panic(err)
-	}
-}

backend/go/llm/gpt2/main.go

@@ -1,23 +0,0 @@
-package main
-
-// Note: this is started internally by LocalAI and a server is allocated for each model
-
-import (
-	"flag"
-
-	transformers "github.com/go-skynet/LocalAI/backend/go/llm/transformers"
-
-	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
-)
-
-var (
-	addr = flag.String("addr", "localhost:50051", "the address to connect to")
-)
-
-func main() {
-	flag.Parse()
-
-	if err := grpc.StartServer(*addr, &transformers.GPT2{}); err != nil {
-		panic(err)
-	}
-}

backend/go/llm/starcoder/main.go

@@ -1,23 +0,0 @@
-package main
-
-// Note: this is started internally by LocalAI and a server is allocated for each model
-
-import (
-	"flag"
-
-	transformers "github.com/go-skynet/LocalAI/backend/go/llm/transformers"
-
-	grpc "github.com/go-skynet/LocalAI/pkg/grpc"
-)
-
-var (
-	addr = flag.String("addr", "localhost:50051", "the address to connect to")
-)
-
-func main() {
-	flag.Parse()
-
-	if err := grpc.StartServer(*addr, &transformers.Starcoder{}); err != nil {
-		panic(err)
-	}
-}

backend/go/llm/transformers/dolly.go

@@ -1,44 +0,0 @@
-package transformers
-
-// This is a wrapper to statisfy the GRPC service interface
-// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
-import (
-	"fmt"
-
-	"github.com/go-skynet/LocalAI/pkg/grpc/base"
-	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
-
-	transformers "github.com/go-skynet/go-ggml-transformers.cpp"
-)
-
-type Dolly struct {
-	base.SingleThread
-
-	dolly *transformers.Dolly
-}
-
-func (llm *Dolly) Load(opts *pb.ModelOptions) error {
-	model, err := transformers.NewDolly(opts.ModelFile)
-	llm.dolly = model
-	return err
-}
-
-func (llm *Dolly) Predict(opts *pb.PredictOptions) (string, error) {
-	return llm.dolly.Predict(opts.Prompt, buildPredictOptions(opts)...)
-}
-
-// fallback to Predict
-func (llm *Dolly) PredictStream(opts *pb.PredictOptions, results chan string) error {
-
-	go func() {
-		res, err := llm.dolly.Predict(opts.Prompt, buildPredictOptions(opts)...)
-
-		if err != nil {
-			fmt.Println("err: ", err)
-		}
-		results <- res
-		close(results)
-	}()
-
-	return nil
-}

backend/go/llm/transformers/falcon.go

@@ -1,43 +0,0 @@
-package transformers
-
-// This is a wrapper to statisfy the GRPC service interface
-// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
-import (
-	"fmt"
-
-	"github.com/go-skynet/LocalAI/pkg/grpc/base"
-	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
-
-	transformers "github.com/go-skynet/go-ggml-transformers.cpp"
-)
-
-type Falcon struct {
-	base.SingleThread
-
-	falcon *transformers.Falcon
-}
-
-func (llm *Falcon) Load(opts *pb.ModelOptions) error {
-	model, err := transformers.NewFalcon(opts.ModelFile)
-	llm.falcon = model
-	return err
-}
-
-func (llm *Falcon) Predict(opts *pb.PredictOptions) (string, error) {
-	return llm.falcon.Predict(opts.Prompt, buildPredictOptions(opts)...)
-}
-
-// fallback to Predict
-func (llm *Falcon) PredictStream(opts *pb.PredictOptions, results chan string) error {
-	go func() {
-		res, err := llm.falcon.Predict(opts.Prompt, buildPredictOptions(opts)...)
-
-		if err != nil {
-			fmt.Println("err: ", err)
-		}
-		results <- res
-		close(results)
-	}()
-
-	return nil
-}

backend/go/llm/transformers/gpt2.go

@@ -1,42 +0,0 @@
-package transformers
-
-// This is a wrapper to statisfy the GRPC service interface
-// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
-import (
-	"fmt"
-
-	"github.com/go-skynet/LocalAI/pkg/grpc/base"
-	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
-
-	transformers "github.com/go-skynet/go-ggml-transformers.cpp"
-)
-
-type GPT2 struct {
-	base.SingleThread
-
-	gpt2 *transformers.GPT2
-}
-
-func (llm *GPT2) Load(opts *pb.ModelOptions) error {
-	model, err := transformers.New(opts.ModelFile)
-	llm.gpt2 = model
-	return err
-}
-
-func (llm *GPT2) Predict(opts *pb.PredictOptions) (string, error) {
-	return llm.gpt2.Predict(opts.Prompt, buildPredictOptions(opts)...)
-}
-
-// fallback to Predict
-func (llm *GPT2) PredictStream(opts *pb.PredictOptions, results chan string) error {
-	go func() {
-		res, err := llm.gpt2.Predict(opts.Prompt, buildPredictOptions(opts)...)
-
-		if err != nil {
-			fmt.Println("err: ", err)
-		}
-		results <- res
-		close(results)
-	}()
-	return nil
-}

backend/go/llm/transformers/gptj.go

@@ -1,42 +0,0 @@
-package transformers
-
-// This is a wrapper to statisfy the GRPC service interface
-// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
-import (
-	"fmt"
-
-	"github.com/go-skynet/LocalAI/pkg/grpc/base"
-	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
-
-	transformers "github.com/go-skynet/go-ggml-transformers.cpp"
-)
-
-type GPTJ struct {
-	base.SingleThread
-
-	gptj *transformers.GPTJ
-}
-
-func (llm *GPTJ) Load(opts *pb.ModelOptions) error {
-	model, err := transformers.NewGPTJ(opts.ModelFile)
-	llm.gptj = model
-	return err
-}
-
-func (llm *GPTJ) Predict(opts *pb.PredictOptions) (string, error) {
-	return llm.gptj.Predict(opts.Prompt, buildPredictOptions(opts)...)
-}
-
-// fallback to Predict
-func (llm *GPTJ) PredictStream(opts *pb.PredictOptions, results chan string) error {
-	go func() {
-		res, err := llm.gptj.Predict(opts.Prompt, buildPredictOptions(opts)...)
-
-		if err != nil {
-			fmt.Println("err: ", err)
-		}
-		results <- res
-		close(results)
-	}()
-	return nil
-}

backend/go/llm/transformers/gptneox.go

@@ -1,42 +0,0 @@
-package transformers
-
-// This is a wrapper to statisfy the GRPC service interface
-// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
-import (
-	"fmt"
-
-	"github.com/go-skynet/LocalAI/pkg/grpc/base"
-	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
-
-	transformers "github.com/go-skynet/go-ggml-transformers.cpp"
-)
-
-type GPTNeoX struct {
-	base.SingleThread
-
-	gptneox *transformers.GPTNeoX
-}
-
-func (llm *GPTNeoX) Load(opts *pb.ModelOptions) error {
-	model, err := transformers.NewGPTNeoX(opts.ModelFile)
-	llm.gptneox = model
-	return err
-}
-
-func (llm *GPTNeoX) Predict(opts *pb.PredictOptions) (string, error) {
-	return llm.gptneox.Predict(opts.Prompt, buildPredictOptions(opts)...)
-}
-
-// fallback to Predict
-func (llm *GPTNeoX) PredictStream(opts *pb.PredictOptions, results chan string) error {
-	go func() {
-		res, err := llm.gptneox.Predict(opts.Prompt, buildPredictOptions(opts)...)
-
-		if err != nil {
-			fmt.Println("err: ", err)
-		}
-		results <- res
-		close(results)
-	}()
-	return nil
-}

backend/go/llm/transformers/mpt.go

@@ -1,42 +0,0 @@
-package transformers
-
-// This is a wrapper to statisfy the GRPC service interface
-// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
-import (
-	"fmt"
-
-	"github.com/go-skynet/LocalAI/pkg/grpc/base"
-	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
-
-	transformers "github.com/go-skynet/go-ggml-transformers.cpp"
-)
-
-type MPT struct {
-	base.SingleThread
-
-	mpt *transformers.MPT
-}
-
-func (llm *MPT) Load(opts *pb.ModelOptions) error {
-	model, err := transformers.NewMPT(opts.ModelFile)
-	llm.mpt = model
-	return err
-}
-
-func (llm *MPT) Predict(opts *pb.PredictOptions) (string, error) {
-	return llm.mpt.Predict(opts.Prompt, buildPredictOptions(opts)...)
-}
-
-// fallback to Predict
-func (llm *MPT) PredictStream(opts *pb.PredictOptions, results chan string) error {
-	go func() {
-		res, err := llm.mpt.Predict(opts.Prompt, buildPredictOptions(opts)...)
-
-		if err != nil {
-			fmt.Println("err: ", err)
-		}
-		results <- res
-		close(results)
-	}()
-	return nil
-}

backend/go/llm/transformers/predict.go

@@ -1,26 +0,0 @@
-package transformers
-
-import (
-	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
-	transformers "github.com/go-skynet/go-ggml-transformers.cpp"
-)
-
-func buildPredictOptions(opts *pb.PredictOptions) []transformers.PredictOption {
-	predictOptions := []transformers.PredictOption{
-		transformers.SetTemperature(float64(opts.Temperature)),
-		transformers.SetTopP(float64(opts.TopP)),
-		transformers.SetTopK(int(opts.TopK)),
-		transformers.SetTokens(int(opts.Tokens)),
-		transformers.SetThreads(int(opts.Threads)),
-	}
-
-	if opts.Batch != 0 {
-		predictOptions = append(predictOptions, transformers.SetBatch(int(opts.Batch)))
-	}
-
-	if opts.Seed != 0 {
-		predictOptions = append(predictOptions, transformers.SetSeed(int(opts.Seed)))
-	}
-
-	return predictOptions
-}

backend/go/llm/transformers/replit.go

@@ -1,42 +0,0 @@
-package transformers
-
-// This is a wrapper to statisfy the GRPC service interface
-// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
-import (
-	"fmt"
-
-	"github.com/go-skynet/LocalAI/pkg/grpc/base"
-	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
-
-	transformers "github.com/go-skynet/go-ggml-transformers.cpp"
-)
-
-type Replit struct {
-	base.SingleThread
-
-	replit *transformers.Replit
-}
-
-func (llm *Replit) Load(opts *pb.ModelOptions) error {
-	model, err := transformers.NewReplit(opts.ModelFile)
-	llm.replit = model
-	return err
-}
-
-func (llm *Replit) Predict(opts *pb.PredictOptions) (string, error) {
-	return llm.replit.Predict(opts.Prompt, buildPredictOptions(opts)...)
-}
-
-// fallback to Predict
-func (llm *Replit) PredictStream(opts *pb.PredictOptions, results chan string) error {
-	go func() {
-		res, err := llm.replit.Predict(opts.Prompt, buildPredictOptions(opts)...)
-
-		if err != nil {
-			fmt.Println("err: ", err)
-		}
-		results <- res
-		close(results)
-	}()
-	return nil
-}

backend/go/llm/transformers/starcoder.go

@@ -1,43 +0,0 @@
-package transformers
-
-// This is a wrapper to statisfy the GRPC service interface
-// It is meant to be used by the main executable that is the server for the specific backend type (falcon, gpt3, etc)
-import (
-	"fmt"
-
-	"github.com/go-skynet/LocalAI/pkg/grpc/base"
-	pb "github.com/go-skynet/LocalAI/pkg/grpc/proto"
-
-	transformers "github.com/go-skynet/go-ggml-transformers.cpp"
-)
-
-type Starcoder struct {
-	base.SingleThread
-
-	starcoder *transformers.Starcoder
-}
-
-func (llm *Starcoder) Load(opts *pb.ModelOptions) error {
-	model, err := transformers.NewStarcoder(opts.ModelFile)
-	llm.starcoder = model
-	return err
-}
-
-func (llm *Starcoder) Predict(opts *pb.PredictOptions) (string, error) {
-	return llm.starcoder.Predict(opts.Prompt, buildPredictOptions(opts)...)
-}
-
-// fallback to Predict
-func (llm *Starcoder) PredictStream(opts *pb.PredictOptions, results chan string) error {
-	go func() {
-		res, err := llm.starcoder.Predict(opts.Prompt, buildPredictOptions(opts)...)
-
-		if err != nil {
-			fmt.Println("err: ", err)
-		}
-		results <- res
-		close(results)
-	}()
-
-	return nil
-}

backend/python/autogptq/Makefile

@@ -1,5 +1,4 @@
 .PHONY: autogptq
 autogptq:
-	@echo "Creating virtual environment..."
-	@conda env create --name autogptq --file autogptq.yml
-	@echo "Virtual environment created."
+	$(MAKE) -C ../common-env/transformers
+

backend/python/autogptq/run.sh

@@ -6,7 +6,7 @@
 export PATH=$PATH:/opt/conda/bin
 
 # Activate conda environment
-source activate autogptq
+source activate transformers
 
 # get the directory where the bash script is located
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

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

@@ -13,3 +13,12 @@ if conda_env_exists "transformers" ; then
 else 
     echo "Virtual environment already exists."
 fi
+
+if [ "$PIP_CACHE_PURGE" = true ] ; then
+    export PATH=$PATH:/opt/conda/bin
+
+    # Activate conda environment
+    source activate transformers
+
+    pip cache purge
+fi

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

@@ -45,7 +45,7 @@ dependencies:
       - fsspec==2023.6.0
       - funcy==2.0
       - grpcio==1.59.0
-      - huggingface-hub==0.16.4
+      - huggingface-hub
       - idna==3.4
       - jinja2==3.1.2
       - jmespath==1.0.1
@@ -70,7 +70,6 @@ dependencies:
       - packaging==23.2
       - pandas
       - peft==0.5.0
-      - git+https://github.com/bigscience-workshop/petals
       - protobuf==4.24.4
       - psutil==5.9.5
       - pyarrow==13.0.0
@@ -85,17 +84,16 @@ dependencies:
       - scipy==1.11.3
       - six==1.16.0
       - sympy==1.12
-      - tokenizers==0.14.0
-      - torch==2.1.0
-      - torchaudio==2.1.0
+      - tokenizers
+      - torch==2.1.2
+      - torchaudio==2.1.2
       - tqdm==4.66.1
-      - transformers==4.34.0
-      - TTS==0.22.0
       - triton==2.1.0
       - typing-extensions==4.8.0
       - tzdata==2023.3
       - urllib3==1.26.17
       - xxhash==3.4.1
+      - auto-gptq==0.6.0
       - yarl==1.9.2
       - soundfile
       - langid
@@ -114,4 +112,7 @@ dependencies:
       - sudachipy
       - sudachidict_core
       - vocos
+      - vllm==0.2.7
+      - transformers>=4.36.0  # Required for Mixtral.
+      - xformers==0.0.23.post1  
 prefix: /opt/conda/envs/transformers

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

@@ -46,7 +46,7 @@ dependencies:
       - fsspec==2023.6.0
       - funcy==2.0
       - grpcio==1.59.0
-      - huggingface-hub==0.16.4
+      - huggingface-hub
       - idna==3.4
       - jinja2==3.1.2
       - jmespath==1.0.1
@@ -59,7 +59,6 @@ dependencies:
       - packaging==23.2
       - pandas
       - peft==0.5.0
-      - git+https://github.com/bigscience-workshop/petals
       - protobuf==4.24.4
       - psutil==5.9.5
       - pyarrow==13.0.0
@@ -74,14 +73,14 @@ dependencies:
       - scipy==1.11.3
       - six==1.16.0
       - sympy==1.12
-      - tokenizers==0.14.0
-      - torch==2.1.0
-      - torchaudio==2.1.0
+      - tokenizers
+      - torch==2.1.2
+      - torchaudio==2.1.2
       - tqdm==4.66.1
-      - transformers==4.34.0
       - triton==2.1.0
       - typing-extensions==4.8.0
       - tzdata==2023.3
+      - auto-gptq==0.6.0
       - urllib3==1.26.17
       - xxhash==3.4.1
       - yarl==1.9.2
@@ -102,4 +101,7 @@ dependencies:
       - sudachipy
       - sudachidict_core
       - vocos
+      - vllm==0.2.7
+      - transformers>=4.36.0  # Required for Mixtral.
+      - xformers==0.0.23.post1  
 prefix: /opt/conda/envs/transformers

backend/python/coqui/coqui_server.py

@@ -21,7 +21,7 @@ _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
 # If MAX_WORKERS are specified in the environment use it, otherwise default to 1
 MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
-COQUI_LANGUAGE = os.environ.get('COQUI_LANGUAGE', 'en')
+COQUI_LANGUAGE = os.environ.get('COQUI_LANGUAGE', None)
 
 # Implement the BackendServicer class with the service methods
 class BackendServicer(backend_pb2_grpc.BackendServicer):
@@ -33,11 +33,18 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
     def LoadModel(self, request, context):
 
         # Get device
-        device = "cuda" if request.CUDA else "cpu"
+        # device = "cuda" if request.CUDA else "cpu"
+        if torch.cuda.is_available():
+            print("CUDA is available", file=sys.stderr)
+            device = "cuda"
+        else:
+            print("CUDA is not available", file=sys.stderr)
+            device = "cpu"
 
         if not torch.cuda.is_available() and request.CUDA:
             return backend_pb2.Result(success=False, message="CUDA is not available")
 
+        self.AudioPath = None
         # List available 🐸TTS models
         print(TTS().list_models())
         if os.path.isabs(request.AudioPath):

backend/python/exllama/Makefile

@@ -1,8 +1,6 @@
 .PHONY: exllama
 exllama:
-	@echo "Creating virtual environment..."
-	@conda env create --name exllama --file exllama.yml
-	@echo "Virtual environment created."
+	$(MAKE) -C ../common-env/transformers
 	bash install.sh
 
 .PHONY: run

backend/python/exllama/install.sh

@@ -5,11 +5,15 @@
 export PATH=$PATH:/opt/conda/bin
 
 # Activate conda environment
-source activate exllama
+source activate transformers
 
 echo $CONDA_PREFIX
 
 
 git clone https://github.com/turboderp/exllama $CONDA_PREFIX/exllama && pushd $CONDA_PREFIX/exllama && pip install -r requirements.txt && popd
 
-cp -rfv $CONDA_PREFIX/exllama/* ./
+cp -rfv $CONDA_PREFIX/exllama/* ./
+
+if [ "$PIP_CACHE_PURGE" = true ] ; then
+    pip cache purge
+fi

backend/python/exllama/run.sh

@@ -6,7 +6,7 @@
 export PATH=$PATH:/opt/conda/bin
 
 # Activate conda environment
-source activate exllama
+source activate transformers
 
 # get the directory where the bash script is located
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

backend/python/exllama2/Makefile

@@ -1,8 +1,6 @@
 .PHONY: exllama2
 exllama2:
-	@echo "Creating virtual environment..."
-	@conda env create --name exllama2 --file exllama2.yml
-	@echo "Virtual environment created."
+	$(MAKE) -C ../common-env/transformers
 	bash install.sh
 
 .PHONY: run

backend/python/exllama2/install.sh

@@ -5,10 +5,14 @@
 export PATH=$PATH:/opt/conda/bin
 
 # Activate conda environment
-source activate exllama2
+source activate transformers
 
 echo $CONDA_PREFIX
 
 git clone https://github.com/turboderp/exllamav2 $CONDA_PREFIX/exllamav2 && pushd $CONDA_PREFIX/exllamav2 && pip install -r requirements.txt && popd
 
-cp -rfv $CONDA_PREFIX/exllamav2/* ./  
+cp -rfv $CONDA_PREFIX/exllamav2/* ./  
+
+if [ "$PIP_CACHE_PURGE" = true ] ; then
+    pip cache purge
+fi

backend/python/exllama2/run.sh

@@ -6,7 +6,7 @@
 export PATH=$PATH:/opt/conda/bin
 
 # Activate conda environment
-source activate exllama2
+source activate transformers
 
 # get the directory where the bash script is located
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

backend/python/mamba/Makefile

@@ -0,0 +1,16 @@
+.PHONY: mamba
+mamba:
+	$(MAKE) -C ../common-env/transformers
+	bash install.sh
+
+.PHONY: run
+run:
+	@echo "Running mamba..."
+	bash run.sh
+	@echo "mamba run."
+
+.PHONY: test
+test:
+	@echo "Testing mamba..."
+	bash test.sh
+	@echo "mamba tested."

backend/python/mamba/README.md

@@ -0,0 +1,5 @@
+# Creating a separate environment for the mamba project
+
+```
+make mamba
+```

backend/python/mamba/backend_mamba.py

@@ -0,0 +1,179 @@
+#!/usr/bin/env python3
+from concurrent import futures
+import time
+import argparse
+import signal
+import sys
+import os
+
+import backend_pb2
+import backend_pb2_grpc
+
+import grpc
+
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+from mamba_ssm.models.mixer_seq_simple import MambaLMHeadModel
+
+_ONE_DAY_IN_SECONDS = 60 * 60 * 24
+
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+MAMBA_CHAT= os.environ.get('MAMBA_CHAT', '1') == '1'
+
+# Implement the BackendServicer class with the service methods
+class BackendServicer(backend_pb2_grpc.BackendServicer):
+    """
+    A gRPC servicer that implements the Backend service defined in backend.proto.
+    """
+    def generate(self,prompt, max_new_tokens):
+        """
+        Generates text based on the given prompt and maximum number of new tokens.
+
+        Args:
+            prompt (str): The prompt to generate text from.
+            max_new_tokens (int): The maximum number of new tokens to generate.
+
+        Returns:
+            str: The generated text.
+        """
+        self.generator.end_beam_search()
+
+        # Tokenizing the input
+        ids = self.generator.tokenizer.encode(prompt)
+
+        self.generator.gen_begin_reuse(ids)
+        initial_len = self.generator.sequence[0].shape[0]
+        has_leading_space = False
+        decoded_text = ''
+        for i in range(max_new_tokens):
+            token = self.generator.gen_single_token()
+            if i == 0 and self.generator.tokenizer.tokenizer.IdToPiece(int(token)).startswith('▁'):
+                has_leading_space = True
+
+            decoded_text = self.generator.tokenizer.decode(self.generator.sequence[0][initial_len:])
+            if has_leading_space:
+                decoded_text = ' ' + decoded_text
+
+            if token.item() == self.generator.tokenizer.eos_token_id:
+                break
+        return decoded_text
+
+    def Health(self, request, context):
+        """
+        Returns a health check message.
+
+        Args:
+            request: The health check request.
+            context: The gRPC context.
+
+        Returns:
+            backend_pb2.Reply: The health check reply.
+        """
+        return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
+
+    def LoadModel(self, request, context):
+        """
+        Loads a language model.
+
+        Args:
+            request: The load model request.
+            context: The gRPC context.
+
+        Returns:
+            backend_pb2.Result: The load model result.
+        """
+        try:
+            tokenizerModel = request.Tokenizer
+            if tokenizerModel == "":
+                tokenizerModel = request.Model
+
+            tokenizer = AutoTokenizer.from_pretrained(tokenizerModel)
+            if MAMBA_CHAT:
+                tokenizer.eos_token = "<|endoftext|>"
+                tokenizer.pad_token = tokenizer.eos_token
+            self.tokenizer = tokenizer
+            self.model = MambaLMHeadModel.from_pretrained(request.Model, device="cuda", dtype=torch.float16)
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+        return backend_pb2.Result(message="Model loaded successfully", success=True)
+
+    def Predict(self, request, context):
+        """
+        Generates text based on the given prompt and sampling parameters.
+
+        Args:
+            request: The predict request.
+            context: The gRPC context.
+
+        Returns:
+            backend_pb2.Result: The predict result.
+        """
+        if request.TopP == 0:
+            request.TopP = 0.9
+
+        max_tokens = request.Tokens
+
+        if request.Tokens == 0:
+            max_tokens = 2000
+
+        # encoded_input = self.tokenizer(request.Prompt)
+        tokens = self.tokenizer(request.Prompt, return_tensors="pt")
+        input_ids = tokens.input_ids.to(device="cuda")
+        out = self.model.generate(input_ids=input_ids, max_length=max_tokens, temperature=request.Temperature,
+                                     top_p=request.TopP, eos_token_id=self.tokenizer.eos_token_id)
+
+        decoded = self.tokenizer.batch_decode(out)
+       
+        generated_text = decoded[0]
+
+        # Remove prompt from response if present
+        if request.Prompt in generated_text:
+            generated_text = generated_text.replace(request.Prompt, "")
+
+        return backend_pb2.Reply(message=bytes(generated_text, encoding='utf-8'))
+
+    def PredictStream(self, request, context):
+        """
+        Generates text based on the given prompt and sampling parameters, and streams the results.
+
+        Args:
+            request: The predict stream request.
+            context: The gRPC context.
+
+        Returns:
+            backend_pb2.Result: The predict stream result.
+        """
+        yield self.Predict(request, context)
+
+def serve(address):
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))
+    backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
+    server.add_insecure_port(address)
+    server.start()
+    print("Server started. Listening on: " + address, file=sys.stderr)
+
+    # Define the signal handler function
+    def signal_handler(sig, frame):
+        print("Received termination signal. Shutting down...")
+        server.stop(0)
+        sys.exit(0)
+
+    # Set the signal handlers for SIGINT and SIGTERM
+    signal.signal(signal.SIGINT, signal_handler)
+    signal.signal(signal.SIGTERM, signal_handler)
+
+    try:
+        while True:
+            time.sleep(_ONE_DAY_IN_SECONDS)
+    except KeyboardInterrupt:
+        server.stop(0)
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Run the gRPC server.")
+    parser.add_argument(
+        "--addr", default="localhost:50051", help="The address to bind the server to."
+    )
+    args = parser.parse_args()
+
+    serve(args.addr)

backend/python/mamba/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/mamba/install.sh

@@ -0,0 +1,21 @@
+#!/bin/bash
+
+##
+## A bash script installs the required dependencies of VALL-E-X and prepares the environment
+export PATH=$PATH:/opt/conda/bin
+
+if [ "$BUILD_TYPE" != "cublas" ]; then
+    echo "[mamba] Attention!!! nvcc is required - skipping installation"
+    exit 0
+fi
+
+# Activate conda environment
+source activate transformers
+
+echo $CONDA_PREFIX
+
+pip install causal-conv1d==1.0.0 mamba-ssm==1.0.1
+
+if [ "$PIP_CACHE_PURGE" = true ] ; then
+    pip cache purge
+fi

backend/python/mamba/run.sh

@@ -0,0 +1,14 @@
+#!/bin/bash
+
+##
+## A bash script wrapper that runs the diffusers 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/backend_mamba.py $@

backend/python/mamba/test.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+##
+## A bash script wrapper that runs the transformers server with conda
+
+# 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 -m unittest $DIR/test_backend_mamba.py

backend/python/mamba/test_backend_mamba.py

@@ -0,0 +1,76 @@
+import unittest
+import subprocess
+import time
+import backend_pb2
+import backend_pb2_grpc
+
+import grpc
+
+import unittest
+import subprocess
+import time
+import grpc
+import backend_pb2_grpc
+import backend_pb2
+
+class TestBackendServicer(unittest.TestCase):
+    """
+    TestBackendServicer is the class that tests the gRPC service.
+
+    This class contains methods to test the startup and shutdown of the gRPC service.
+    """
+    def setUp(self):
+        self.service = subprocess.Popen(["python", "backend_vllm.py", "--addr", "localhost:50051"])
+        time.sleep(10)
+
+    def tearDown(self) -> None:
+        self.service.terminate()
+        self.service.wait()
+
+    def test_server_startup(self):
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.Health(backend_pb2.HealthMessage())
+                self.assertEqual(response.message, b'OK')
+        except Exception as err:
+            print(err)
+            self.fail("Server failed to start")
+        finally:
+            self.tearDown()
+    def test_load_model(self):
+        """
+        This method tests if the model is loaded successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.LoadModel(backend_pb2.ModelOptions(Model="facebook/opt-125m"))
+                self.assertTrue(response.success)
+                self.assertEqual(response.message, "Model loaded successfully")
+        except Exception as err:
+            print(err)
+            self.fail("LoadModel service failed")
+        finally:
+            self.tearDown()
+
+    def test_text(self):
+        """
+        This method tests if the embeddings are generated successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.LoadModel(backend_pb2.ModelOptions(Model="facebook/opt-125m"))
+                self.assertTrue(response.success)
+                req = backend_pb2.PredictOptions(Prompt="The capital of France is")
+                resp = stub.Predict(req)
+                self.assertIsNotNone(resp.message)
+        except Exception as err:
+            print(err)
+            self.fail("text service failed")
+        finally:
+            self.tearDown()

backend/python/petals/Makefile

@@ -1,6 +1,8 @@
 .PHONY: petals
 petals:
-	$(MAKE) -C ../common-env/transformers
+	@echo "Creating virtual environment..."
+	@conda env create --name petals --file petals.yml
+	@echo "Virtual environment created."
 
 .PHONY: run
 run:

backend/python/petals/run.sh

@@ -5,14 +5,16 @@
 
 export PATH=$PATH:/opt/conda/bin
 
+CONDA_ENV=petals
+
 # Activate conda environment
 # if source is available use it, or use conda
 #
 if [ -f /opt/conda/bin/activate ]; then
-    source activate transformers
+    source activate $CONDA_ENV
 else
     eval "$(conda shell.bash hook)"
-    conda activate transformers
+    conda activate $CONDA_ENV
 fi
 
 # get the directory where the bash script is located

backend/python/petals/test.sh

@@ -3,7 +3,16 @@
 ## A bash script wrapper that runs the transformers server with conda
 
 # Activate conda environment
-source activate transformers
+CONDA_ENV=petals
+# Activate conda environment
+# if source is available use it, or use conda
+#
+if [ -f /opt/conda/bin/activate ]; then
+    source activate $CONDA_ENV
+else
+    eval "$(conda shell.bash hook)"
+    conda activate $CONDA_ENV
+fi
 
 # get the directory where the bash script is located
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

backend/python/transformers/transformers_server.py

@@ -15,8 +15,8 @@ import backend_pb2_grpc
 
 import grpc
 import torch
-
-from transformers import AutoTokenizer, AutoModel
+import torch.cuda
+from transformers import AutoTokenizer, AutoModel, AutoModelForCausalLM, set_seed
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
@@ -68,16 +68,19 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         """
         model_name = request.Model
         try:
-            self.model = AutoModel.from_pretrained(model_name, trust_remote_code=True) # trust_remote_code is needed to use the encode method with embeddings models like jinai-v2
-            self.tokenizer = AutoTokenizer.from_pretrained(model_name)
+            if request.Type == "AutoModelForCausalLM":
+                self.model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True)
+            else:
+                self.model = AutoModel.from_pretrained(model_name, trust_remote_code=True)
 
-            if request.CUDA:
+            self.tokenizer = AutoTokenizer.from_pretrained(model_name)
+            self.CUDA = False
+
+            if request.CUDA or torch.cuda.is_available():
                 try:
-                    # TODO: also tensorflow, make configurable
-                    import torch.cuda
-                    if torch.cuda.is_available():
-                        print("Loading model", model_name, "to CUDA.", file=sys.stderr)
-                        self.model = self.model.to("cuda")
+                    print("Loading model", model_name, "to CUDA.", file=sys.stderr)
+                    self.model = self.model.to("cuda")
+                    self.CUDA = True
                 except Exception as err:
                     print("Not using CUDA:", err, file=sys.stderr)
         except Exception as err:
@@ -98,6 +101,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             An EmbeddingResult object that contains the calculated embeddings.
         """
 
+        set_seed(request.Seed)
         # Tokenize input
         max_length = 512
         if request.Tokens != 0:
@@ -113,6 +117,51 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         print("Embeddings:", sentence_embeddings, file=sys.stderr)
         return backend_pb2.EmbeddingResult(embeddings=sentence_embeddings)
 
+    def Predict(self, request, context):
+        """
+        Generates text based on the given prompt and sampling parameters.
+
+        Args:
+            request: The predict request.
+            context: The gRPC context.
+
+        Returns:
+            backend_pb2.Reply: The predict result.
+        """
+        set_seed(request.Seed)
+        if request.TopP == 0:
+            request.TopP = 0.9
+
+        max_tokens = 200
+        if request.Tokens > 0:
+            max_tokens = request.Tokens
+
+        inputs = self.tokenizer(request.Prompt, return_tensors="pt").input_ids
+        if self.CUDA:
+            inputs = inputs.to("cuda")
+
+        outputs = self.model.generate(inputs,max_new_tokens=max_tokens, temperature=request.Temperature, top_p=request.TopP)
+
+        generated_text = self.tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]
+        # Remove prompt from response if present
+        if request.Prompt in generated_text:
+            generated_text = generated_text.replace(request.Prompt, "")
+
+        return backend_pb2.Reply(message=bytes(generated_text, encoding='utf-8'))
+
+    def PredictStream(self, request, context):
+        """
+        Generates text based on the given prompt and sampling parameters, and streams the results.
+
+        Args:
+            request: The predict stream request.
+            context: The gRPC context.
+
+        Returns:
+            backend_pb2.Result: The predict stream result.
+        """
+        yield self.Predict(request, context)
+
 
 def serve(address):
     server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))

backend/python/vall-e-x/install.sh

@@ -12,4 +12,8 @@ echo $CONDA_PREFIX
 
 git clone https://github.com/Plachtaa/VALL-E-X.git $CONDA_PREFIX/vall-e-x && pushd $CONDA_PREFIX/vall-e-x && git checkout -b build $SHA && pip install -r requirements.txt && popd
 
-cp -rfv $CONDA_PREFIX/vall-e-x/* ./
+cp -rfv $CONDA_PREFIX/vall-e-x/* ./
+
+if [ "$PIP_CACHE_PURGE" = true ] ; then
+    pip cache purge
+fi

backend/python/vllm/Makefile

@@ -1,8 +1,6 @@
 .PHONY: vllm
 vllm:
-	@echo "Creating virtual environment..."
-	@conda env create --name vllm --file vllm.yml
-	@echo "Virtual environment created."
+	$(MAKE) -C ../common-env/transformers
 
 .PHONY: run
 run:

backend/python/vllm/backend_vllm.py

@@ -97,12 +97,16 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             context: The gRPC context.
 
         Returns:
-            backend_pb2.Result: The predict result.
+            backend_pb2.Reply: The predict result.
         """
         if request.TopP == 0:
             request.TopP = 0.9
 
-        sampling_params = SamplingParams(temperature=request.Temperature, top_p=request.TopP)
+        max_tokens = 200
+        if request.Tokens > 0:
+            max_tokens = request.Tokens
+
+        sampling_params = SamplingParams(max_tokens=max_tokens, temperature=request.Temperature, top_p=request.TopP)
         outputs = self.llm.generate([request.Prompt], sampling_params)
 
         generated_text = outputs[0].outputs[0].text
@@ -110,7 +114,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         if request.Prompt in generated_text:
             generated_text = generated_text.replace(request.Prompt, "")
 
-        return backend_pb2.Result(message=bytes(generated_text, encoding='utf-8'))
+        return backend_pb2.Reply(message=bytes(generated_text, encoding='utf-8'))
 
     def PredictStream(self, request, context):
         """
@@ -123,11 +127,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         Returns:
             backend_pb2.Result: The predict stream result.
         """
-        # Implement PredictStream RPC
-        #for reply in some_data_generator():
-        #    yield reply
-        # Not implemented yet
-        return self.Predict(request, context)
+        yield self.Predict(request, context)
 
 def serve(address):
     server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS))

backend/python/vllm/run.sh

@@ -6,7 +6,7 @@
 export PATH=$PATH:/opt/conda/bin
 
 # Activate conda environment
-source activate vllm
+source activate transformers
 
 # get the directory where the bash script is located
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

backend/python/vllm/test.sh

@@ -3,7 +3,7 @@
 ## A bash script wrapper that runs the transformers server with conda
 
 # Activate conda environment
-source activate vllm
+source activate transformers
 
 # get the directory where the bash script is located
 DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

backend/python/vllm/vllm.yml

@@ -1,99 +0,0 @@
-name: vllm
-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:
-      - aiosignal==1.3.1
-      - anyio==3.7.1
-      - attrs==23.1.0
-      - certifi==2023.7.22
-      - charset-normalizer==3.3.0
-      - click==8.1.7
-      - cmake==3.27.6
-      - fastapi==0.103.2
-      - filelock==3.12.4
-      - frozenlist==1.4.0
-      - fsspec==2023.9.2
-      - grpcio==1.59.0
-      - h11==0.14.0
-      - httptools==0.6.0
-      - huggingface-hub==0.17.3
-      - idna==3.4
-      - jinja2==3.1.2
-      - jsonschema==4.19.1
-      - jsonschema-specifications==2023.7.1
-      - lit==17.0.2
-      - markupsafe==2.1.3
-      - mpmath==1.3.0
-      - msgpack==1.0.7
-      - networkx==3.1
-      - ninja==1.11.1
-      - numpy==1.26.0
-      - nvidia-cublas-cu11==11.10.3.66
-      - nvidia-cuda-cupti-cu11==11.7.101
-      - nvidia-cuda-nvrtc-cu11==11.7.99
-      - nvidia-cuda-runtime-cu11==11.7.99
-      - nvidia-cudnn-cu11==8.5.0.96
-      - nvidia-cufft-cu11==10.9.0.58
-      - nvidia-curand-cu11==10.2.10.91
-      - nvidia-cusolver-cu11==11.4.0.1
-      - nvidia-cusparse-cu11==11.7.4.91
-      - nvidia-nccl-cu11==2.14.3
-      - nvidia-nvtx-cu11==11.7.91
-      - packaging==23.2
-      - pandas==2.1.1
-      - protobuf==4.24.4
-      - psutil==5.9.5
-      - pyarrow==13.0.0
-      - pydantic==1.10.13
-      - python-dateutil==2.8.2
-      - python-dotenv==1.0.0
-      - pytz==2023.3.post1
-      - pyyaml==6.0.1
-      - ray==2.7.0
-      - referencing==0.30.2
-      - regex==2023.10.3
-      - requests==2.31.0
-      - rpds-py==0.10.4
-      - safetensors==0.4.0
-      - sentencepiece==0.1.99
-      - six==1.16.0
-      - sniffio==1.3.0
-      - starlette==0.27.0
-      - sympy==1.12
-      - tokenizers==0.14.1
-      - torch==2.0.1
-      - tqdm==4.66.1
-      - transformers==4.34.0
-      - triton==2.0.0
-      - typing-extensions==4.8.0
-      - tzdata==2023.3
-      - urllib3==2.0.6
-      - uvicorn==0.23.2
-      - uvloop==0.17.0
-      - vllm==0.2.0
-      - watchfiles==0.20.0
-      - websockets==11.0.3
-      - xformers==0.0.22
-prefix: /opt/conda/envs/vllm

docs/assets/jsconfig.json

@@ -0,0 +1,11 @@
+{
+ "compilerOptions": {
+  "baseUrl": ".",
+  "paths": {
+   "*": [
+    "../../../../.cache/hugo_cache/modules/filecache/modules/pkg/mod/github.com/gohugoio/hugo-mod-jslibs-dist/popperjs/v2@v2.21100.20000/package/dist/cjs/popper.js/*",
+    "../../../../.cache/hugo_cache/modules/filecache/modules/pkg/mod/github.com/twbs/bootstrap@v5.3.2+incompatible/js/*"
+   ]
+  }
+ }
+}

docs/config.toml

@@ -1,133 +1,178 @@
-# this is a required setting for this theme to appear on https://themes.gohugo.io/
-# change this to a value appropriate for you; if your site is served from a subdirectory
-# set it like "https://example.com/mysite/"
 baseURL = "https://localai.io/"
+languageCode = "en-GB"
+contentDir = "content"
+enableEmoji = true
+enableGitInfo = true # N.B. .GitInfo does not currently function with git submodule content directories
 
-# canonicalization will only be used for the sitemap.xml and index.xml files;
-# if set to false, a site served from a subdirectory will generate wrong links
-# inside of the above mentioned files; if you serve the page from the servers root
-# you are free to set the value to false as recommended by the official Hugo documentation
-canonifyURLs = true # true -> all relative URLs would instead be canonicalized using baseURL
-# required value to serve this page from a webserver AND the file system;
-# if you don't want to serve your page from the file system, you can also set this value
-# to false
-relativeURLs = true # true -> rewrite all relative URLs to be relative to the current content
-# if you set uglyURLs to false, this theme will append 'index.html' to any branch bundle link
-# so your page can be also served from the file system; if you don't want that,
-# set disableExplicitIndexURLs=true in the [params] section
-uglyURLs = false     # true -> basic/index.html -> basic.html
+defaultContentLanguage = 'en'
 
-# the directory where Hugo reads the themes from; this is specific to your
-# installation and most certainly needs be deleted or changed
-#themesdir = "../.."
-# yeah, well, obviously a mandatory setting for your site, if you want to
-# use this theme ;-)
-theme = "hugo-theme-relearn"
-
-# the main language of this site; also an automatic pirrrate translation is
-# available in this showcase
-languageCode = "en"
-
-# make sure your defaultContentLanguage is the first one in the [languages]
-# array below, as the theme needs to make assumptions on it
-defaultContentLanguage = "en"
-
-# the site's title of this showcase; you should change this ;-)
-title = "LocalAI Documentation"
-
-# We disable this for testing the exampleSite; you must do so too
-# if you want to use the themes parameter disableGeneratorVersion=true;
-# otherwise Hugo will create a generator tag on your home page
-disableHugoGeneratorInject = true
-
-[outputs]
-  # add JSON to the home to support Lunr search; This is a mandatory setting
-  # for the search functionality
-  # add PRINT to home, section and page to activate the feature to print whole
-  # chapters
-  home = ["HTML", "RSS", "PRINT", "SEARCH", "SEARCHPAGE"]
-  section = ["HTML", "RSS", "PRINT"]
-  page = ["HTML", "RSS", "PRINT"]
 
 [markup]
-  [markup.highlight]
-    # if `guessSyntax = true`, there will be no unstyled code even if no language
-    # was given BUT Mermaid and Math codefences will not work anymore! So this is a
-    # mandatory setting for your site if you want to use Mermaid or Math codefences
-    guessSyntax = true
+  defaultMarkdownHandler = "goldmark"
+  [markup.tableOfContents]
+      endLevel = 3
+      startLevel = 1
+  [markup.goldmark]
+    [markup.goldmark.renderer]
+      unsafe = true # https://jdhao.github.io/2019/12/29/hugo_html_not_shown/
+  # [markup.highlight]
+  #   codeFences = false # disables Hugo's default syntax highlighting
+  # [markup.goldmark.parser]
+  #   [markup.goldmark.parser.attribute]
+  #     block = true
+  #     title = true
 
-    # here in this showcase we use our own modified chroma syntax highlightning style
-    # which is imported in theme-relearn-light.css / theme-relearn-dark.css;
-    # if you want to use a predefined style instead:
-    # - remove the following `noClasses`
-    # - set the following `style` to a predefined style name
-    # - remove the `@import` of the self-defined chroma stylesheet from your CSS files
-    #   (here eg.: theme-relearn-light.css / theme-relearn-dark.css)
-    noClasses = false
-    style = "tango"
 
-  [markup.goldmark.renderer]
-    # activated for this showcase to use HTML and JavaScript; decide on your own needs;
-    # if in doubt, remove this line
-    unsafe = true
 
-# allows `hugo server` to display this showcase in IE11; this is used for testing, as we
-# are still supporting IE11 - although with degraded experience; if you don't care about
-# `hugo server` or browsers of ancient times, fell free to remove this whole block
-[server]
-  [[server.headers]]
-    for = "**.html"
-    [server.headers.values]
-       X-UA-Compatible = "IE=edge"
+[params]
+
+  google_fonts = [
+    ["Inter", "300, 400, 600, 700"],
+    ["Fira Code", "500, 700"]
+  ]
+
+  sans_serif_font = "Inter"     # Default is System font
+  secondary_font  = "Inter"     # Default is System font
+  mono_font       = "Fira Code" # Default is System font
+
+    [params.footer]
+        copyright = "© 2023-2024 <a href='https://mudler.pm' target=_blank>Ettore Di Giacinto</a>"
+        version = true # includes git commit info
+
+    [params.social]
+        github = "mudler/LocalAI"        # YOUR_GITHUB_ID or YOUR_GITHUB_URL
+        twitter = "LocalAI_API"       # YOUR_TWITTER_ID
+        dicord = "uJAeKSAGDy"
+        # instagram = "colinwilson"     # YOUR_INSTAGRAM_ID
+        rss = true                    # show rss icon with link
+
+    [params.docs] # Parameters for the /docs 'template'
+
+        logo = "https://github.com/go-skynet/LocalAI/assets/2420543/0966aa2a-166e-4f99-a3e5-6c915fc997dd"
+        logo_text = "LocalAI"
+        title           = "LocalAI documentation"           # default html title for documentation pages/sections
+
+        pathName        = "docs"                            # path name for documentation site | default "docs"
+
+        # themeColor      = "cyan"                            # (optional) - Set theme accent colour. Options include: blue (default), green, red, yellow, emerald, cardinal, magenta, cyan
+
+        darkMode        = true                                # enable dark mode option? default false
+
+        prism           = true                                # enable syntax highlighting via Prism
+
+        prismTheme      = "solarized-light"                           # (optional) - Set theme for PrismJS. Options include: lotusdocs (default), solarized-light, twilight, lucario
+
+        # gitinfo
+        repoURL         = "https://github.com/mudler/LocalAI"  # Git repository URL for your site [support for GitHub, GitLab, and BitBucket]
+        repoBranch      = "master"
+        editPage        = true                                # enable 'Edit this page' feature - default false
+        lastMod         = true                                # enable 'Last modified' date on pages - default false
+        lastModRelative = true                                # format 'Last modified' time as relative - default true
+
+        sidebarIcons    = true                                # enable sidebar icons? default false
+        breadcrumbs     = true                                # default is true
+        backToTop       = true                                # enable back-to-top button? default true
+
+        # ToC
+        toc             = true                                # enable table of contents? default is true
+        tocMobile       = true                                # enable table of contents in mobile view? default is true
+        scrollSpy       = true                                # enable scrollspy on ToC? default is true
+
+        # front matter
+        descriptions    = true                                # enable front matter descriptions under content title?
+        titleIcon       = true                                # enable front matter icon title prefix? default is false
+
+        # content navigation
+        navDesc         = true                                # include front matter descriptions in Prev/Next navigation cards
+        navDescTrunc    = 30                                  # Number of characters by which to truncate the Prev/Next descriptions
+
+        listDescTrunc   = 100                                 # Number of characters by which to truncate the list card description
+
+        # Link behaviour
+        intLinkTooltip  = true                                # Enable a tooltip for internal links that displays info about the destination? default false
+        # extLinkNewTab   = false                             # Open external links in a new Tab? default true
+        # logoLinkURL = ""                                    # Set a custom URL destination for the top header logo link.
+
+    [params.flexsearch] # Parameters for FlexSearch
+        enabled             = true
+        # tokenize            = "full"
+        # optimize            = true
+        # cache               = 100
+        # minQueryChar        = 3 # default is 0 (disabled)
+        # maxResult           = 5 # default is 5
+        # searchSectionsIndex = []
+
+    [params.docsearch] # Parameters for DocSearch
+        # appID     = "" # Algolia Application ID
+        # apiKey    = "" # Algolia Search-Only API (Public) Key
+        # indexName = "" # Index Name to perform search on (or set env variable HUGO_PARAM_DOCSEARCH_indexName)
+
+    [params.analytics] # Parameters for Analytics (Google, Plausible)
+        # plausibleURL    = "/docs/s" # (or set via env variable HUGO_PARAM_ANALYTICS_plausibleURL)
+        # plausibleAPI    = "/docs/s" # optional - (or set via env variable HUGO_PARAM_ANALYTICS_plausibleAPI)
+        # plausibleDomain = ""      # (or set via env variable HUGO_PARAM_ANALYTICS_plausibleDomain)
+
+    # [params.feedback]
+    #     enabled = true
+    #     emoticonTpl = true
+    #     eventDest = ["plausible","google"]
+    #     emoticonEventName = "Feedback"
+    #     positiveEventName = "Positive Feedback"
+    #     negativeEventName = "Negative Feedback"
+    #     positiveFormTitle = "What did you like?"
+    #     negativeFormTitle = "What went wrong?"
+    #     successMsg = "Thank you for helping to improve Lotus Docs' documentation!"
+    #     errorMsg = "Sorry! There was an error while attempting to submit your feedback!"
+    #     positiveForm = [
+    #       ["Accurate", "Accurately describes the feature or option."],
+    #       ["Solved my problem", "Helped me resolve an issue."],
+    #       ["Easy to understand", "Easy to follow and comprehend."],
+    #       ["Something else"]
+    #     ]
+    #     negativeForm = [
+    #       ["Inaccurate", "Doesn't accurately describe the feature or option."],
+    #       ["Couldn't find what I was looking for", "Missing important information."],
+    #       ["Hard to understand", "Too complicated or unclear."],
+    #       ["Code sample errors", "One or more code samples are incorrect."],
+    #       ["Something else"]
+    #     ]
+
+[menu]
+ [[menu.primary]]
+    name  = "Docs"
+    url = "docs/"
+    identifier = "docs"
+    weight = 10
+[[menu.primary]]
+    name = "Discord"
+    url = "https://discord.gg/uJAeKSAGDy"
+    identifier = "discord"
+    weight = 20
 
-# showcase of the menu shortcuts; you can use relative URLs linking
-# to your content or use fully-quallified URLs to link outside of
-# your project
 [languages]
   [languages.en]
     title = "LocalAI documentation"
-    weight = 1
     languageName = "English"
-    [languages.en.params]
-      landingPageName = "<i class='fas fa-home'></i> Home"
-  [[languages.en.menu.shortcuts]]
-    name = "<i class='fas fa-home'></i> Home"
-    url = "/"
-    weight = 1
-  [[languages.en.menu.shortcuts]]
-    name = "<i class='fab fa-fw fa-github'></i> GitHub repo"
-    identifier = "ds"
-    url = "https://github.com/go-skynet/LocalAI"
     weight = 10
+#  [languages.fr]
+#    title = "LocalAI documentation"
+#    languageName = "Français"
+#    contentDir = "content/fr"
+#    weight = 20
+#  [languages.de]
+#    title = "LocalAI documentation"
+#    languageName = "Deutsch"
+#    contentDir = "content/de"
+#    weight = 30
 
-  [[languages.en.menu.shortcuts]]
-    name = "<i class='fas fa-fw fa-camera'></i> Examples"
-    url = "https://github.com/go-skynet/LocalAI/tree/master/examples/"
-    weight = 11
 
-  [[languages.en.menu.shortcuts]]
-    name = "<i class='fas fa-fw fa-images'></i> Model Gallery"
-    url = "https://github.com/go-skynet/model-gallery"
-    weight = 12
 
-  [[languages.en.menu.shortcuts]]
-    name = "<i class='fas fa-fw fa-download'></i> Container images"
-    url = "https://quay.io/repository/go-skynet/local-ai"
-    weight = 20
-  #[[languages.en.menu.shortcuts]]
-  #  name = "<i class='fas fa-fw fa-bullhorn'></i> Credits"
-  #  url = "more/credits/"
-  #  weight = 30
-
-  [[languages.en.menu.shortcuts]]
-    name = "<i class='fas fa-fw fa-tags'></i> Releases"
-    url = "https://github.com/go-skynet/LocalAI/releases"
-    weight = 40
 
 
 # mounts are only needed in this showcase to access the publicly available screenshots;
 # remove this section if you don't need further mounts
 [module]
+  replacements = "github.com/colinwilson/lotusdocs -> lotusdocs"
   [[module.mounts]]
     source = 'archetypes'
     target = 'archetypes'
@@ -152,30 +197,11 @@ disableHugoGeneratorInject = true
   [[module.mounts]]
     source = 'static'
     target = 'static'
-
-
-# settings specific to this theme's features; choose to your likings and
-# consult this documentation for explaination
-[params]
-  editURL = "https://github.com/mudler/LocalAI/edit/master/docs/content/"
-  description = "Documentation for LocalAI"
-  author = "Ettore Di Giacinto"
-  showVisitedLinks = true
-  collapsibleMenu = true
-  disableBreadcrumb = false
-  disableInlineCopyToClipBoard = true
-  disableNextPrev = false
-  disableLandingPageButton = true
-  breadcrumbSeparator = ">"
-  titleSeparator = "::"
-  themeVariant = [ "auto", "relearn-bright", "relearn-light", "relearn-dark", "learn", "neon", "blue", "green", "red" ]
-  themeVariantAuto = [ "relearn-light", "relearn-dark" ]
-  disableSeoHiddenPages = true
-  # this is to index search for your native language in other languages, too (eg.
-  # pir in this showcase)
-  additionalContentLanguage = [ "en" ]
-  # this is for the stylesheet generator to allow for interactivity in Mermaid
-  # graphs; you usually will not need it and you should remove this for
-  # security reasons
-  mermaidInitialize = "{ \"securityLevel\": \"loose\" }"
-  mermaidZoom = true
+    # uncomment line below for temporary local development of module
+    # or when using a 'theme' as a git submodule
+  [[module.imports]]
+    path = "github.com/colinwilson/lotusdocs"
+    disable = false
+  [[module.imports]]
+    path = "github.com/gohugoio/hugo-mod-bootstrap-scss/v5"
+    disable = false

docs/content/advanced/development.md

@@ -1,37 +0,0 @@
-
-+++
-disableToc = false
-title = "Development documentation"
-weight = 7
-+++
-
-{{% notice note %}}
-
-This section is for developers and contributors. If you are looking for the user documentation, this is not the right place!
-
-{{% /notice %}}
-
-This section will collect how-to, notes and development documentation
-
-## Contributing
-
-We use conventional commits and semantic versioning. Please follow the [conventional commits](https://www.conventionalcommits.org/en/v1.0.0/) specification when writing commit messages.
-
-## Creating a gRPC backend
-
-LocalAI backends are `gRPC` servers.
-
-In order to create a new backend you need:
-
-- If there are changes required to the protobuf code, modify the [proto](https://github.com/go-skynet/LocalAI/blob/master/pkg/grpc/proto/backend.proto) file and re-generate the code with `make protogen`.
-- Modify the `Makefile` to add your new backend and re-generate the client code with `make protogen` if necessary.
-- Create a new `gRPC` server in `extra/grpc` if it's not written in go: [link](https://github.com/go-skynet/LocalAI/tree/master/extra/grpc), and create the specific implementation.
-    - Golang `gRPC` servers should be added in the [pkg/backend](https://github.com/go-skynet/LocalAI/tree/master/pkg/backend) directory given their type. See [piper](https://github.com/go-skynet/LocalAI/blob/master/pkg/backend/tts/piper.go) as an example.
-    - Golang servers needs a respective `cmd/grpc` binary that must be created too, see also [cmd/grpc/piper](https://github.com/go-skynet/LocalAI/tree/master/cmd/grpc/piper) as an example, update also the Makefile accordingly to build the binary during build time.
-- Update the Dockerfile: if the backend is written in another language, update the `Dockerfile` default *EXTERNAL_GRPC_BACKENDS* variable by listing the new binary [link](https://github.com/go-skynet/LocalAI/blob/c2233648164f67cdb74dd33b8d46244e14436ab3/Dockerfile#L14).
-
-Once you are done, you can either re-build `LocalAI` with your backend or you can try it out by running the `gRPC` server manually and specifying the host and IP to LocalAI with `--external-grpc-backends` or using (`EXTERNAL_GRPC_BACKENDS` environment variable, comma separated list of `name:host:port` tuples, e.g. `my-awesome-backend:host:port`):
-
-```bash
-./local-ai --debug --external-grpc-backends "my-awesome-backend:host:port" ...
-```

docs/content/docs/advanced/_index.en.md

@@ -0,0 +1,11 @@
+---
+weight: 20
+title: "Advanced"
+description: "Advanced usage"
+icon: science
+lead: ""
+date: 2020-10-06T08:49:15+00:00
+lastmod: 2020-10-06T08:49:15+00:00
+draft: false
+images: []
+---

docs/content/docs/advanced/advanced-usage.md

@@ -1,8 +1,9 @@
 
 +++
 disableToc = false
-title = "Advanced"
-weight = 6
+title = "Advanced usage"
+weight = 21
+url = '/advanced'
 +++
 
 ### Advanced configuration with YAML files
@@ -309,7 +310,7 @@ prompt_cache_all: true
 
 By default LocalAI will try to autoload the model by trying all the backends. This might work for most of models, but some of the backends are NOT configured to autoload.
 
-The available backends are listed in the [model compatibility table]({{%relref "model-compatibility" %}}).
+The available backends are listed in the [model compatibility table]({{%relref "docs/reference/compatibility-table" %}}).
 
 In order to specify a backend for your models, create a model config file in your `models` directory specifying the backend:
 
@@ -343,6 +344,19 @@ Or a remote URI:
 ./local-ai --debug --external-grpc-backends "my-awesome-backend:host:port"
 ```
 
+For example, to start vllm manually after compiling LocalAI (also assuming running the command from the root of the repository):
+
+```bash
+./local-ai --external-grpc-backends "vllm:$PWD/backend/python/vllm/run.sh"
+```
+
+Note that first is is necessary to create the conda environment with:
+
+```bash
+make -C backend/python/vllm
+```
+
+
 ### Environment variables
 
 When LocalAI runs in a container,
@@ -419,11 +433,11 @@ RUN PATH=$PATH:/opt/conda/bin make -C backend/python/diffusers
 ENV EXTERNAL_GRPC_BACKENDS="diffusers:/build/backend/python/diffusers/run.sh"
 ```
 
-{{% notice note %}}
+{{% alert note %}}
 
 You can specify remote external backends or path to local files. The syntax is `backend-name:/path/to/backend` or `backend-name:host:port`.
 
-{{% /notice %}}
+{{% /alert %}}
 
 #### In runtime
 

docs/content/docs/advanced/fine-tuning.md

@@ -2,12 +2,12 @@
 +++
 disableToc = false
 title = "Fine-tuning LLMs for text generation"
-weight = 3
+weight = 22
 +++
 
-{{% notice note %}}
+{{% alert note %}}
 Section under construction
-{{% /notice %}}
+{{% /alert %}}
 
 This section covers how to fine-tune a language model for text generation and consume it in LocalAI.
 
@@ -23,7 +23,7 @@ Fine-tuning a language model is a process that requires a lot of computational p
 
 Currently LocalAI doesn't support the fine-tuning endpoint as LocalAI but there are are [plans](https://github.com/mudler/LocalAI/issues/596) to support that. For the time being a guide is proposed here to give a simple starting point on how to fine-tune a model and use it with LocalAI (but also with llama.cpp).
 
-There is an e2e example of fine-tuning a LLM model to use with [LocalAI](https://github/mudler/LocalAI) written by [@mudler](https://github.com/mudler) available [here](https://github.com/mudler/LocalAI/tree/master/examples/e2e-fine-tuning/).
+There is an e2e example of fine-tuning a LLM model to use with [LocalAI](https://github.com/mudler/LocalAI) written by [@mudler](https://github.com/mudler) available [here](https://github.com/mudler/LocalAI/tree/master/examples/e2e-fine-tuning/).
 
 The steps involved are:
 

docs/content/docs/faq.md

@@ -2,7 +2,9 @@
 +++
 disableToc = false
 title = "FAQ"
-weight = 9
+weight = 24
+icon = "quiz"
+url = "/faq/"
 +++
 
 ## Frequently asked questions
@@ -12,25 +14,13 @@ Here are answers to some of the most common questions.
 
 ### How do I get models? 
 
-<details>
-
 Most gguf-based models should work, but newer models may require additions to the API. If a model doesn't work, please feel free to open up issues. However, be cautious about downloading models from the internet and directly onto your machine, as there may be security vulnerabilities in lama.cpp or ggml that could be maliciously exploited. Some models can be found on Hugging Face: https://huggingface.co/models?search=gguf, or models from gpt4all are compatible too: https://github.com/nomic-ai/gpt4all.
 
-</details>
-
 ### What's the difference with Serge, or XXX?
 
-
-<details>
-
 LocalAI is a multi-model solution that doesn't focus on a specific model type (e.g., llama.cpp or alpaca.cpp), and it handles all of these internally for faster inference,  easy to set up locally and deploy to Kubernetes.
 
-</details>
-
-
-### Everything is slow, how come?
-
-<details>
+### Everything is slow, how is it possible?
 
 There are few situation why this could occur. Some tips are:
 - Don't use HDD to store your models. Prefer SSD over HDD. In case you are stuck with HDD, disable `mmap` in the model config file so it loads everything in memory.
@@ -38,61 +28,31 @@ There are few situation why this could occur. Some tips are:
 - Run LocalAI with `DEBUG=true`. This gives more information, including stats on the token inference speed.
 - Check that you are actually getting an output: run a simple curl request with `"stream": true` to see how fast the model is responding. 
 
-</details>
-
 ### Can I use it with a Discord bot, or XXX?
 
-<details>
-
 Yes! If the client uses OpenAI and supports setting a different base URL to send requests to, you can use the LocalAI endpoint. This allows to use this with every application that was supposed to work with OpenAI, but without changing the application!
 
-</details>
-
-
 ### Can this leverage GPUs? 
 
-<details>
-
-There is partial GPU support, see build instructions above.
-
-</details>
+There is GPU support, see {{%relref "docs/features/GPU-acceleration" %}}.
 
 ### Where is the webUI? 
 
-<details> 
-There is the availability of localai-webui and chatbot-ui in the examples section and can be setup as per the instructions. However as LocalAI is an API you can already plug it into existing projects that provides are UI interfaces to OpenAI's APIs. There are several already on github, and should be compatible with LocalAI already (as it mimics the OpenAI API)
-
-</details>
+There is the availability of localai-webui and chatbot-ui in the examples section and can be setup as per the instructions. However as LocalAI is an API you can already plug it into existing projects that provides are UI interfaces to OpenAI's APIs. There are several already on Github, and should be compatible with LocalAI already (as it mimics the OpenAI API)
 
 ### Does it work with AutoGPT? 
 
-<details>
-
 Yes, see the [examples](https://github.com/go-skynet/LocalAI/tree/master/examples/)!
 
-</details>
-
 ### How can I troubleshoot when something is wrong?
 
-<details>
-
 Enable the debug mode by setting `DEBUG=true` in the environment variables. This will give you more information on what's going on.
 You can also specify `--debug` in the command line.
 
-</details>
-
 ### I'm getting 'invalid pitch' error when running with CUDA, what's wrong?
 
-<details>
-
 This typically happens when your prompt exceeds the context size. Try to reduce the prompt size, or increase the context size.
 
-</details>
-
 ### I'm getting a 'SIGILL' error, what's wrong?
 
-<details>
-
-Your CPU probably does not have support for certain instructions that are compiled by default in the pre-built binaries. If you are running in a container, try setting `REBUILD=true` and disable the CPU instructions that are not compatible with your CPU. For instance: `CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF" make build`
-  
-</details>
+Your CPU probably does not have support for certain instructions that are compiled by default in the pre-built binaries. If you are running in a container, try setting `REBUILD=true` and disable the CPU instructions that are not compatible with your CPU. For instance: `CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF" make build`

docs/content/docs/features/GPU-acceleration.md

@@ -1,21 +1,59 @@
-
 +++
 disableToc = false
 title = "⚡ GPU acceleration"
-weight = 2
+weight = 9
+url = "/features/gpu-acceleration/"
 +++
 
-{{% notice note %}}
+{{% alert context="warning" %}}
 Section under construction
-{{% /notice %}}
+{{% /alert %}}
 
 This section contains instruction on how to use LocalAI with GPU acceleration.
 
-{{% notice note %}}
-For accelleration for AMD or Metal HW there are no specific container images, see the [build]({{%relref "build/#acceleration" %}})
-{{% /notice %}}
+{{% alert icon="⚡" context="warning" %}}
+For accelleration for AMD or Metal HW there are no specific container images, see the [build]({{%relref "docs/getting-started/build#Acceleration" %}})
+{{% /alert %}}
 
-### CUDA(NVIDIA) acceleration
+
+## Model configuration
+
+Depending on the model architecture and backend used, there might be different ways to enable GPU acceleration. It is required to configure the model you intend to use with a YAML config file. For example, for `llama.cpp` workloads a configuration file might look like this (where `gpu_layers` is the number of layers to offload to the GPU):
+
+```yaml
+name: my-model-name
+# Default model parameters
+parameters:
+  # Relative to the models path
+  model: llama.cpp-model.ggmlv3.q5_K_M.bin
+
+context_size: 1024
+threads: 1
+
+f16: true # enable with GPU acceleration
+gpu_layers: 22 # GPU Layers (only used when built with cublas)
+
+```
+
+For diffusers instead, it might look like this instead:
+
+```yaml
+name: stablediffusion
+parameters:
+  model: toonyou_beta6.safetensors
+backend: diffusers
+step: 30
+f16: true
+diffusers:
+  pipeline_type: StableDiffusionPipeline
+  cuda: true
+  enable_parameters: "negative_prompt,num_inference_steps,clip_skip"
+  scheduler_type: "k_dpmpp_sde"
+```
+
+## CUDA(NVIDIA) acceleration
+
+### Requirements
 
 Requirement: nvidia-container-toolkit (installation instructions [1](https://www.server-world.info/en/note?os=Ubuntu_22.04&p=nvidia&f=2) [2](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html))
 
@@ -72,37 +110,21 @@ llama_model_load_internal: total VRAM used: 1598 MB
 llama_init_from_file: kv self size  =  512.00 MB
 ```
 
-#### Model configuration
+## Intel acceleration (sycl)
 
-Depending on the model architecture and backend used, there might be different ways to enable GPU acceleration. It is required to configure the model you intend to use with a YAML config file. For example, for `llama.cpp` workloads a configuration file might look like this (where `gpu_layers` is the number of layers to offload to the GPU):
+#### Requirements
 
-```yaml
-name: my-model-name
-# Default model parameters
-parameters:
-  # Relative to the models path
-  model: llama.cpp-model.ggmlv3.q5_K_M.bin
+Requirement: [Intel oneAPI Base Toolkit](https://software.intel.com/content/www/us/en/develop/tools/oneapi/base-toolkit/download.html)
 
-context_size: 1024
-threads: 1
+To use SYCL, use the images with the `sycl-f16` or `sycl-f32` tag, for example `{{< version >}}-sycl-f32-core`, `{{< version >}}-sycl-f16-ffmpeg-core`, ...
 
-f16: true # enable with GPU acceleration
-gpu_layers: 22 # GPU Layers (only used when built with cublas)
+The image list is on [quay](https://quay.io/repository/go-skynet/local-ai?tab=tags).
 
+### Notes
+
+In addition to the commands to run LocalAI normally, you need to specify `--device /dev/dri` to docker, for example:
+
+```bash
+docker run --rm -ti --device /dev/dri -p 8080:8080 -e DEBUG=true -e MODELS_PATH=/models -e THREADS=1 -v $PWD/models:/models quay.io/go-skynet/local-ai:{{< version >}}-sycl-f16-ffmpeg-core
 ```
 
-For diffusers instead, it might look like this instead:
-
-```yaml
-name: stablediffusion
-parameters:
-  model: toonyou_beta6.safetensors
-backend: diffusers
-step: 30
-f16: true
-diffusers:
-  pipeline_type: StableDiffusionPipeline
-  cuda: true
-  enable_parameters: "negative_prompt,num_inference_steps,clip_skip"
-  scheduler_type: "k_dpmpp_sde"
-```

docs/content/docs/features/_index.en.md

@@ -0,0 +1,8 @@
+
++++
+disableToc = false
+title = "Features"
+weight = 8
+icon = "feature_search"
+url = "/features/"
++++

docs/content/docs/features/audio-to-text.md

@@ -1,10 +1,13 @@
 +++
 disableToc = false
 title = "🔈 Audio to text"
-weight = 2
+weight = 16
+url = "/features/audio-to-text/"
 +++
 
-The transcription endpoint allows to convert audio files to text. The endpoint is based on [whisper.cpp](https://github.com/ggerganov/whisper.cpp), a C++ library for audio transcription. The endpoint supports the audio formats supported by `ffmpeg`.
+Audio to text models are models that can generate text from an audio file.
+
+The transcription endpoint allows to convert audio files to text. The endpoint is based on [whisper.cpp](https://github.com/ggerganov/whisper.cpp), a C++ library for audio transcription. The endpoint input supports all the audio formats supported by `ffmpeg`.
 
 ## Usage
 

docs/content/docs/features/constrained_grammars.md

@@ -2,20 +2,21 @@
 +++
 disableToc = false
 title = "✍️ Constrained grammars"
-weight = 6
+weight = 15
+url = "/features/constrained_grammars/"
 +++
 
 The chat endpoint accepts an additional `grammar` parameter which takes a [BNF defined grammar](https://en.wikipedia.org/wiki/Backus%E2%80%93Naur_form).
 
 This allows the LLM to constrain the output to a user-defined schema, allowing to generate `JSON`, `YAML`, and everything that can be defined with a BNF grammar.
 
-{{% notice note %}}
-This feature works only with models compatible with the [llama.cpp](https://github.com/ggerganov/llama.cpp) backend (see also [Model compatibility]({{%relref "model-compatibility" %}})). For details on how it works, see the upstream PRs: https://github.com/ggerganov/llama.cpp/pull/1773, https://github.com/ggerganov/llama.cpp/pull/1887
-{{% /notice %}}
+{{% alert note %}}
+This feature works only with models compatible with the [llama.cpp](https://github.com/ggerganov/llama.cpp) backend (see also [Model compatibility]({{%relref "docs/reference/compatibility-table" %}})). For details on how it works, see the upstream PRs: https://github.com/ggerganov/llama.cpp/pull/1773, https://github.com/ggerganov/llama.cpp/pull/1887
+{{% /alert %}}
 
 ## Setup
 
-Follow the setup instructions from the [LocalAI functions]({{%relref "features/openai-functions" %}}) page.
+Follow the setup instructions from the [LocalAI functions]({{%relref "docs/features/openai-functions" %}}) page.
 
 ## 💡 Usage example
 

docs/content/docs/features/embeddings.md

@@ -2,7 +2,8 @@
 +++
 disableToc = false
 title = "🧠 Embeddings"
-weight = 2
+weight = 13
+url = "/features/embeddings/"
 +++
 
 LocalAI supports generating embeddings for text or list of tokens.
@@ -73,7 +74,7 @@ parameters:
 
 The `sentencetransformers` backend uses Python [sentence-transformers](https://github.com/UKPLab/sentence-transformers). For a list of all pre-trained models available see here: https://github.com/UKPLab/sentence-transformers#pre-trained-models
 
-{{% notice note %}}
+{{% alert note %}}
 
 - The `sentencetransformers` backend is an optional backend of LocalAI and uses Python. If you are running `LocalAI` from the containers you are good to go and should be already configured for use.
 - If you are running `LocalAI` manually you must install the python dependencies (`make prepare-extra-conda-environments`). This requires `conda` to be installed.
@@ -82,7 +83,7 @@ The `sentencetransformers` backend uses Python [sentence-transformers](https://g
 - The `sentencetransformers` backend does support only embeddings of text, and not of tokens. If you need to embed tokens you can use the `bert` backend or `llama.cpp`.
 - No models are required to be downloaded before using the `sentencetransformers` backend. The models will be downloaded automatically the first time the API is used.
 
-{{% /notice %}}
+{{% /alert %}}
 
 ## Llama.cpp embeddings
 

docs/content/docs/features/gpt-vision.md

@@ -2,13 +2,10 @@
 +++
 disableToc = false
 title = "🆕 GPT Vision"
-weight = 2
+weight = 14
+url = "/features/gpt-vision/"
 +++
 
-{{% notice note %}}
-Available only on `master` builds
-{{% /notice %}}
-
 LocalAI supports understanding images by using [LLaVA](https://llava.hliu.cc/), and implements the [GPT Vision API](https://platform.openai.com/docs/guides/vision) from OpenAI.
 
 ![llava](https://github.com/mudler/LocalAI/assets/2420543/cb0a0897-3b58-4350-af66-e6f4387b58d3)
@@ -27,4 +24,4 @@ curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/jso
 
 ### Setup
 
-To setup the LLaVa models, follow the full example in the [configuration examples](https://github.com/mudler/LocalAI/blob/master/examples/configurations/README.md#llava).
+To setup the LLaVa models, follow the full example in the [configuration examples](https://github.com/mudler/LocalAI/blob/master/examples/configurations/README.md#llava).

docs/content/docs/features/image-generation.md

@@ -2,13 +2,14 @@
 +++
 disableToc = false
 title = "🎨 Image generation"
-weight = 2
+weight = 12
+url = "/features/image-generation/"
 +++
 
 ![anime_girl](https://github.com/go-skynet/LocalAI/assets/2420543/8aaca62a-e864-4011-98ae-dcc708103928)
 (Generated with [AnimagineXL](https://huggingface.co/Linaqruf/animagine-xl))
 
-LocalAI supports generating images with Stable diffusion, running on CPU using a C++ implementation, [Stable-Diffusion-NCNN](https://github.com/EdVince/Stable-Diffusion-NCNN) ([binding](https://github.com/mudler/go-stable-diffusion)) and [🧨 Diffusers]({{%relref "model-compatibility/diffusers" %}}).
+LocalAI supports generating images with Stable diffusion, running on CPU using C++ and Python implementations.
 
 ## Usage
 
@@ -35,7 +36,9 @@ curl http://localhost:8080/v1/images/generations -H "Content-Type: application/j
 }'
 ```
 
-## stablediffusion-cpp
+## Backends
+
+### stablediffusion-cpp
 
 | mode=0                                                                                                                | mode=1 (winograd/sgemm)                                                                                                                |
 |------------------------------------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------------|
@@ -45,7 +48,7 @@ curl http://localhost:8080/v1/images/generations -H "Content-Type: application/j
 
 Note: image generator supports images up to 512x512. You can use other tools however to upscale the image, for instance: https://github.com/upscayl/upscayl.
 
-### Setup
+#### Setup
 
 Note: In order to use the `images/generation` endpoint with the `stablediffusion` C++ backend, you need to build LocalAI with `GO_TAGS=stablediffusion`. If you are using the container images, it is already enabled.
 
@@ -128,11 +131,14 @@ models
 
 {{< /tabs >}}
 
-## Diffusers
+### Diffusers
 
-This is an extra backend - in the container is already available and there is nothing to do for the setup.
+[Diffusers](https://huggingface.co/docs/diffusers/index) is the go-to library for state-of-the-art pretrained diffusion models for generating images, audio, and even 3D structures of molecules. LocalAI has a diffusers backend which allows image generation using the `diffusers` library.
 
-### Model setup
+![anime_girl](https://github.com/go-skynet/LocalAI/assets/2420543/8aaca62a-e864-4011-98ae-dcc708103928)
+(Generated with [AnimagineXL](https://huggingface.co/Linaqruf/animagine-xl))
+
+#### Model setup
 
 The models will be downloaded the first time you use the backend from `huggingface` automatically.
 
@@ -150,3 +156,198 @@ diffusers:
   cuda: false # Enable for GPU usage (CUDA)
   scheduler_type: euler_a
 ```
+
+#### Dependencies
+
+This is an extra backend - in the container is already available and there is nothing to do for the setup. Do not use *core* images (ending with `-core`). If you are building manually, see the [build instructions]({{%relref "docs/getting-started/build" %}}).
+
+#### Model setup
+
+The models will be downloaded the first time you use the backend from `huggingface` automatically.
+
+Create a model configuration file in the `models` directory, for instance to use `Linaqruf/animagine-xl` with CPU:
+
+```yaml
+name: animagine-xl
+parameters:
+  model: Linaqruf/animagine-xl
+backend: diffusers
+cuda: true
+f16: true
+diffusers:
+  scheduler_type: euler_a
+```
+
+#### Local models
+
+You can also use local models, or modify some parameters like `clip_skip`, `scheduler_type`, for instance:
+
+```yaml
+name: stablediffusion
+parameters:
+  model: toonyou_beta6.safetensors
+backend: diffusers
+step: 30
+f16: true
+cuda: true
+diffusers:
+  pipeline_type: StableDiffusionPipeline
+  enable_parameters: "negative_prompt,num_inference_steps,clip_skip"
+  scheduler_type: "k_dpmpp_sde"
+  cfg_scale: 8
+  clip_skip: 11
+```
+
+#### Configuration parameters
+
+The following parameters are available in the configuration file:
+
+| Parameter | Description | Default |
+| --- | --- | --- |
+| `f16` | Force the usage of `float16` instead of `float32` | `false` |
+| `step` | Number of steps to run the model for | `30` |
+| `cuda` | Enable CUDA acceleration | `false` |
+| `enable_parameters` | Parameters to enable for the model | `negative_prompt,num_inference_steps,clip_skip` |
+| `scheduler_type` | Scheduler type | `k_dpp_sde` |
+| `cfg_scale` | Configuration scale | `8` |
+| `clip_skip` | Clip skip | None |
+| `pipeline_type` | Pipeline type | `AutoPipelineForText2Image` |
+
+There are available several types of schedulers:
+
+| Scheduler | Description |
+| --- | --- |
+| `ddim` | DDIM |
+| `pndm` | PNDM |
+| `heun` | Heun |
+| `unipc` | UniPC |
+| `euler` | Euler |
+| `euler_a` | Euler a |
+| `lms` | LMS |
+| `k_lms` | LMS Karras |
+| `dpm_2` | DPM2 |
+| `k_dpm_2` | DPM2 Karras |
+| `dpm_2_a` | DPM2 a |
+| `k_dpm_2_a` | DPM2 a Karras |
+| `dpmpp_2m` | DPM++ 2M |
+| `k_dpmpp_2m` | DPM++ 2M Karras |
+| `dpmpp_sde` | DPM++ SDE |
+| `k_dpmpp_sde` | DPM++ SDE Karras |
+| `dpmpp_2m_sde` | DPM++ 2M SDE |
+| `k_dpmpp_2m_sde` | DPM++ 2M SDE Karras |
+
+Pipelines types available:
+
+| Pipeline type | Description |
+| --- | --- |
+| `StableDiffusionPipeline` | Stable diffusion pipeline |
+| `StableDiffusionImg2ImgPipeline` | Stable diffusion image to image pipeline |
+| `StableDiffusionDepth2ImgPipeline` | Stable diffusion depth to image pipeline |
+| `DiffusionPipeline` | Diffusion pipeline |
+| `StableDiffusionXLPipeline` | Stable diffusion XL pipeline |
+
+#### Usage
+
+#### Text to Image
+Use the `image` generation endpoint with the `model` name from the configuration file:
+
+```bash
+curl http://localhost:8080/v1/images/generations \
+    -H "Content-Type: application/json" \
+    -d '{
+      "prompt": "<positive prompt>|<negative prompt>", 
+      "model": "animagine-xl", 
+      "step": 51,
+      "size": "1024x1024" 
+    }'
+```
+
+#### Image to Image
+
+https://huggingface.co/docs/diffusers/using-diffusers/img2img
+
+An example model (GPU):
+```yaml
+name: stablediffusion-edit
+parameters:
+  model: nitrosocke/Ghibli-Diffusion
+backend: diffusers
+step: 25
+cuda: true
+f16: true
+diffusers:
+  pipeline_type: StableDiffusionImg2ImgPipeline
+  enable_parameters: "negative_prompt,num_inference_steps,image"
+```
+
+```bash
+IMAGE_PATH=/path/to/your/image
+(echo -n '{"file": "'; base64 $IMAGE_PATH; echo '", "prompt": "a sky background","size": "512x512","model":"stablediffusion-edit"}') |
+curl -H "Content-Type: application/json" -d @-  http://localhost:8080/v1/images/generations
+```
+
+#### Depth to Image
+
+https://huggingface.co/docs/diffusers/using-diffusers/depth2img
+
+```yaml
+name: stablediffusion-depth
+parameters:
+  model: stabilityai/stable-diffusion-2-depth
+backend: diffusers
+step: 50
+# Force CPU usage
+f16: true
+cuda: true
+diffusers:
+  pipeline_type: StableDiffusionDepth2ImgPipeline
+  enable_parameters: "negative_prompt,num_inference_steps,image"
+  cfg_scale: 6
+```
+
+```bash
+(echo -n '{"file": "'; base64 ~/path/to/image.jpeg; echo '", "prompt": "a sky background","size": "512x512","model":"stablediffusion-depth"}') |
+curl -H "Content-Type: application/json" -d @-  http://localhost:8080/v1/images/generations
+```
+
+#### img2vid
+
+
+```yaml
+name: img2vid
+parameters:
+  model: stabilityai/stable-video-diffusion-img2vid
+backend: diffusers
+step: 25
+# Force CPU usage
+f16: true
+cuda: true
+diffusers:
+  pipeline_type: StableVideoDiffusionPipeline
+```
+
+```bash
+(echo -n '{"file": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/svd/rocket.png?download=true","size": "512x512","model":"img2vid"}') |
+curl -H "Content-Type: application/json" -X POST -d @- http://localhost:8080/v1/images/generations
+```
+
+#### txt2vid
+
+```yaml
+name: txt2vid
+parameters:
+  model: damo-vilab/text-to-video-ms-1.7b
+backend: diffusers
+step: 25
+# Force CPU usage
+f16: true
+cuda: true
+diffusers:
+  pipeline_type: VideoDiffusionPipeline
+  cuda: true
+```
+
+```bash
+(echo -n '{"prompt": "spiderman surfing","size": "512x512","model":"txt2vid"}') |
+curl -H "Content-Type: application/json" -X POST -d @- http://localhost:8080/v1/images/generations
+```

docs/content/docs/features/model-gallery.md

@@ -2,7 +2,9 @@
 +++
 disableToc = false
 title = "🖼️ Model gallery"
-weight = 7
+
+weight = 18
+url = '/models'
 +++
 
 <h1 align="center">
@@ -15,13 +17,13 @@ The model gallery is a (experimental!) collection of models configurations for [
 
 LocalAI to ease out installations of models provide a way to preload models on start and downloading and installing them in runtime. You can install models manually by copying them over the `models` directory, or use the API to configure, download and verify the model assets for you. As the UI is still a work in progress, you will find here the documentation about the API Endpoints.
 
-{{% notice note %}}
+{{% alert note %}}
 The models in this gallery are not directly maintained by LocalAI. If you find a model that is not working, please open an issue on the model gallery repository.
-{{% /notice %}}
+{{% /alert %}}
 
-{{% notice note %}}
+{{% alert note %}}
 GPT and text generation models might have a license which is not permissive for commercial use or might be questionable or without any license at all. Please check the model license before using it. The official gallery contains only open licensed models.
-{{% /notice %}}
+{{% /alert %}}
 
 ## Useful Links and resources
 
@@ -48,7 +50,7 @@ GALLERIES=[{"name":"model-gallery", "url":"github:go-skynet/model-gallery/index.
 
 where `github:go-skynet/model-gallery/index.yaml` will be expanded automatically to `https://raw.githubusercontent.com/go-skynet/model-gallery/main/index.yaml`.
 
-{{% notice note %}}
+{{% alert note %}}
 
 As this feature is experimental, you need to run `local-ai` with a list of `GALLERIES`. Currently there are two galleries:
 
@@ -63,19 +65,19 @@ GALLERIES=[{"name":"model-gallery", "url":"github:go-skynet/model-gallery/index.
 
 If running with `docker-compose`, simply edit the `.env` file and uncomment the `GALLERIES` variable, and add the one you want to use.
 
-{{% /notice %}}
+{{% /alert %}}
 
-{{% notice note %}}
+{{% alert note %}}
 You might not find all the models in this gallery. Automated CI updates the gallery automatically. You can find however most of the models on huggingface (https://huggingface.co/), generally it should be available `~24h` after upload.
 
 By under any circumstances LocalAI and any developer is not responsible for the models in this gallery, as CI is just indexing them and providing a convenient way to install with an automatic configuration with a consistent API. Don't install models from authors you don't trust, and, check the appropriate license for your use case. Models are automatically indexed and hosted on huggingface (https://huggingface.co/). For any issue with the models, please open an issue on the model gallery repository if it's a LocalAI misconfiguration, otherwise refer to the huggingface repository. If you think a model should not be listed, please reach to us and we will remove it from the gallery.
-{{% /notice %}}
+{{% /alert %}}
 
-{{% notice note %}}
+{{% alert note %}}
 
 There is no documentation yet on how to build a gallery or a repository - but you can find an example in the [model-gallery](https://github.com/go-skynet/model-gallery) repository.
 
-{{% /notice %}}
+{{% /alert %}}
 
 
 ### List Models
@@ -117,7 +119,7 @@ where:
 - `bert-embeddings` is the model name in the gallery
   (read its [config here](https://github.com/go-skynet/model-gallery/blob/main/bert-embeddings.yaml)).
 
-{{% notice note %}}
+{{% alert note %}}
 If the `huggingface` model gallery is enabled (it's enabled by default),
 and the model has an entry in the model gallery's associated YAML config
 (for `huggingface`, see [`model-gallery/huggingface.yaml`](https://github.com/go-skynet/model-gallery/blob/main/huggingface.yaml)),
@@ -132,7 +134,7 @@ curl $LOCALAI/models/apply -H "Content-Type: application/json" -d '{
 ```
 
 Note that the `id` can be used similarly when pre-loading models at start.
-{{% /notice %}}
+{{% /alert %}}
 
 
 ## How to install a model (without a gallery)
@@ -217,7 +219,7 @@ YAML:
 
 </details>
 
-{{% notice note %}}
+{{% alert note %}}
 
 You can find already some open licensed models in the [model gallery](https://github.com/go-skynet/model-gallery).
 
@@ -241,7 +243,7 @@ curl $LOCALAI/models/apply -H "Content-Type: application/json" -d '{
 
 </details>
 
-{{% /notice %}}
+{{% /alert %}}
 
 ## Installing a model with a different name
 

docs/content/docs/features/openai-functions.md

@@ -2,7 +2,8 @@
 +++
 disableToc = false
 title = "🔥 OpenAI functions"
-weight = 2
+weight = 17
+url = "/features/openai-functions/"
 +++
 
 LocalAI supports running OpenAI functions with `llama.cpp` compatible models.
@@ -67,13 +68,13 @@ response = openai.ChatCompletion.create(
 # ...
 ```
 
-{{% notice note %}}
+{{% alert note %}}
 When running the python script, be sure to:
 
 - Set `OPENAI_API_KEY` environment variable to a random string (the OpenAI api key is NOT required!)
 - Set `OPENAI_API_BASE` to point to your LocalAI service, for example `OPENAI_API_BASE=http://localhost:8080`
 
-{{% /notice %}}
+{{% /alert %}}
 
 ## Advanced
 

docs/content/docs/features/text-generation.md

@@ -0,0 +1,264 @@
+
++++
+disableToc = false
+title = "📖 Text generation (GPT)"
+weight = 10
+url = "/features/text-generation/"
++++
+
+LocalAI supports generating text with GPT with `llama.cpp` and other backends (such as `rwkv.cpp` as ) see also the [Model compatibility]({{%relref "docs/reference/compatibility-table" %}}) for an up-to-date list of the supported model families.
+
+Note:
+
+- You can also specify the model name as part of the OpenAI token.
+- If only one model is available, the API will use it for all the requests.
+
+## API Reference
+
+### Chat completions
+
+https://platform.openai.com/docs/api-reference/chat
+
+For example, to generate a chat completion, you can send a POST request to the `/v1/chat/completions` endpoint with the instruction as the request body:
+
+```bash
+curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
+  "model": "ggml-koala-7b-model-q4_0-r2.bin",
+  "messages": [{"role": "user", "content": "Say this is a test!"}],
+  "temperature": 0.7
+}'
+```
+
+Available additional parameters: `top_p`, `top_k`, `max_tokens`
+
+### Edit completions
+
+https://platform.openai.com/docs/api-reference/edits
+
+To generate an edit completion you can send a POST request to the `/v1/edits` endpoint with the instruction as the request body:
+
+```bash
+curl http://localhost:8080/v1/edits -H "Content-Type: application/json" -d '{
+  "model": "ggml-koala-7b-model-q4_0-r2.bin",
+  "instruction": "rephrase",
+  "input": "Black cat jumped out of the window",
+  "temperature": 0.7
+}'
+```
+
+Available additional parameters: `top_p`, `top_k`, `max_tokens`.
+
+### Completions
+
+https://platform.openai.com/docs/api-reference/completions
+
+To generate a completion, you can send a POST request to the `/v1/completions` endpoint with the instruction as per the request body:
+
+```bash
+curl http://localhost:8080/v1/completions -H "Content-Type: application/json" -d '{
+  "model": "ggml-koala-7b-model-q4_0-r2.bin",
+  "prompt": "A long time ago in a galaxy far, far away",
+  "temperature": 0.7
+}'
+```
+
+Available additional parameters: `top_p`, `top_k`, `max_tokens`
+
+### List models
+
+You can list all the models available with:
+
+```bash
+curl http://localhost:8080/v1/models
+```
+
+## Backends
+
+### AutoGPTQ
+
+[AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ) is an easy-to-use LLMs quantization package with user-friendly apis, based on GPTQ algorithm.
+
+#### Prerequisites
+
+This is an extra backend - in the container images is already available and there is nothing to do for the setup.
+
+If you are building LocalAI locally, you need to install [AutoGPTQ manually](https://github.com/PanQiWei/AutoGPTQ#quick-installation).
+
+
+#### Model setup
+
+The models are automatically downloaded from `huggingface` if not present the first time. It is possible to define models via `YAML` config file, or just by querying the endpoint with the `huggingface` repository model name. For example, create a `YAML` config file in `models/`:
+
+```
+name: orca
+backend: autogptq
+model_base_name: "orca_mini_v2_13b-GPTQ-4bit-128g.no-act.order"
+parameters:
+  model: "TheBloke/orca_mini_v2_13b-GPTQ"
+# ...
+```
+
+Test with:
+
+```bash
+curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{                                                                                                         
+   "model": "orca",
+   "messages": [{"role": "user", "content": "How are you?"}],
+   "temperature": 0.1
+ }'
+```
+### RWKV
+
+A full example on how to run a rwkv model is in the [examples](https://github.com/go-skynet/LocalAI/tree/master/examples/rwkv).
+
+Note: rwkv models needs to specify the backend `rwkv` in the YAML config files and have an associated tokenizer along that needs to be provided with it:
+
+```
+36464540 -rw-r--r--  1 mudler mudler 1.2G May  3 10:51 rwkv_small
+36464543 -rw-r--r--  1 mudler mudler 2.4M May  3 10:51 rwkv_small.tokenizer.json
+```
+
+### llama.cpp
+
+[llama.cpp](https://github.com/ggerganov/llama.cpp) is a popular port of Facebook's LLaMA model in C/C++.
+
+{{% alert note %}}
+
+The `ggml` file format has been deprecated. If you are using `ggml` models and you are configuring your model with a YAML file, specify, use the `llama-ggml` backend instead. If you are relying in automatic detection of the model, you should be fine. For `gguf` models, use the `llama` backend. The go backend is deprecated as well but still available as `go-llama`. The go backend supports still features not available in the mainline: speculative sampling and embeddings.
+
+{{% /alert %}}
+
+#### Features
+
+The `llama.cpp` model supports the following features:
+- [📖 Text generation (GPT)]({{%relref "docs/features/text-generation" %}})
+- [🧠 Embeddings]({{%relref "docs/features/embeddings" %}})
+- [🔥 OpenAI functions]({{%relref "docs/features/openai-functions" %}})
+- [✍️ Constrained grammars]({{%relref "docs/features/constrained_grammars" %}})
+
+#### Setup
+
+LocalAI supports `llama.cpp` models out of the box. You can use the `llama.cpp` model in the same way as any other model. 
+
+##### Manual setup
+
+It is sufficient to copy the `ggml` or `gguf` model files in the `models` folder. You can refer to the model in the `model` parameter in the API calls.
+
+[You can optionally create an associated YAML]({{%relref "docs/advanced" %}}) model config file to tune the model's parameters or apply a template to the prompt.
+
+Prompt templates are useful for models that are fine-tuned towards a specific prompt. 
+
+##### Automatic setup
+
+LocalAI supports model galleries which are indexes of models. For instance, the huggingface gallery contains a large curated index of models from the huggingface model hub for `ggml` or `gguf` models.
+
+For instance, if you have the galleries enabled and LocalAI already running, you can just start chatting with models in huggingface by running:
+
+```bash
+curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
+     "model": "TheBloke/WizardLM-13B-V1.2-GGML/wizardlm-13b-v1.2.ggmlv3.q2_K.bin",
+     "messages": [{"role": "user", "content": "Say this is a test!"}],
+     "temperature": 0.1
+   }'
+```
+
+LocalAI will automatically download and configure the model in the `model` directory.
+
+Models can be also preloaded or downloaded on demand. To learn about model galleries, check out the [model gallery documentation]({{%relref "docs/features/model-gallery" %}}).
+
+#### YAML configuration
+
+To use the `llama.cpp` backend, specify `llama` as the backend in the YAML file:
+
+```yaml
+name: llama
+backend: llama
+parameters:
+  # Relative to the models path
+  model: file.gguf.bin
+```
+
+In the example above we specify `llama` as the backend to restrict loading `gguf` models only. 
+
+For instance, to use the `llama-ggml` backend for `ggml` models:
+
+```yaml
+name: llama
+backend: llama-ggml
+parameters:
+  # Relative to the models path
+  model: file.ggml.bin
+```
+
+#### Reference
+
+- [llama](https://github.com/ggerganov/llama.cpp)
+- [binding](https://github.com/go-skynet/go-llama.cpp)
+
+
+### exllama/2
+
+[Exllama](https://github.com/turboderp/exllama) is a "A more memory-efficient rewrite of the HF transformers implementation of Llama for use with quantized weights". Both `exllama` and `exllama2` are supported.
+
+#### Model setup
+
+Download the model as a folder inside the `model ` directory and create a YAML file specifying the `exllama` backend. For instance with the `TheBloke/WizardLM-7B-uncensored-GPTQ` model:
+
+```
+$ git lfs install
+$ cd models && git clone https://huggingface.co/TheBloke/WizardLM-7B-uncensored-GPTQ
+$ ls models/                                                                 
+.keep                        WizardLM-7B-uncensored-GPTQ/ exllama.yaml
+$ cat models/exllama.yaml                                                     
+name: exllama
+parameters:
+  model: WizardLM-7B-uncensored-GPTQ
+backend: exllama
+# Note: you can also specify "exllama2" if it's an exllama2 model here
+# ...
+```
+
+Test with:
+
+```bash
+curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{                                                                                                         
+   "model": "exllama",
+   "messages": [{"role": "user", "content": "How are you?"}],
+   "temperature": 0.1
+ }'
+```
+
+### vLLM
+
+[vLLM](https://github.com/vllm-project/vllm) is a fast and easy-to-use library for LLM inference.
+
+LocalAI has a built-in integration with vLLM, and it can be used to run models. You can check out `vllm` performance [here](https://github.com/vllm-project/vllm#performance).
+
+#### Setup
+
+Create a YAML file for the model you want to use with `vllm`.
+
+To setup a model, you need to just specify the model name in the YAML config file:
+```yaml
+name: vllm
+backend: vllm
+parameters:
+    model: "facebook/opt-125m"
+
+# Decomment to specify a quantization method (optional)
+# quantization: "awq"
+```
+
+The backend will automatically download the required files in order to run the model.
+
+
+#### Usage
+
+Use the `completions` endpoint by specifying the `vllm` backend:
+```
+curl http://localhost:8080/v1/completions -H "Content-Type: application/json" -d '{   
+   "model": "vllm",
+   "prompt": "Hello, my name is",
+   "temperature": 0.1, "top_p": 0.1
+ }'
+```

docs/content/docs/features/text-to-audio.md

@@ -0,0 +1,159 @@
+
++++
+disableToc = false
+title = "🗣 Text to audio (TTS)"
+weight = 11
+url = "/features/text-to-audio/"
++++
+
+The `/tts` endpoint can be used to generate speech from text.
+
+## Usage
+
+Input: `input`, `model`
+
+For example, to generate an audio file, you can send a POST request to the `/tts` endpoint with the instruction as the request body:
+
+```bash
+curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{
+  "input": "Hello world",
+  "model": "tts"
+}'
+```
+
+Returns an `audio/wav` file.
+
+
+## Backends
+
+### 🐸 Coqui
+
+Required: Don't use `LocalAI` images ending with the `-core` tag,. Python dependencies are required in order to use this backend.
+
+Coqui works without any configuration, to test it, you can run the following curl command:
+
+```
+    curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{         
+        "backend": "coqui",
+        "model": "tts_models/en/ljspeech/glow-tts",
+        "input":"Hello, this is a test!"
+        }'
+```
+
+### Bark
+
+[Bark](https://github.com/suno-ai/bark) allows to generate audio from text prompts.
+
+This is an extra backend - in the container is already available and there is nothing to do for the setup.
+
+#### Model setup
+
+There is nothing to be done for the model setup. You can already start to use bark. The models will be downloaded the first time you use the backend.
+
+#### Usage
+
+Use the `tts` endpoint by specifying the `bark` backend:
+
+```
+curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{         
+     "backend": "bark",
+     "input":"Hello!"
+   }' | aplay
+```
+
+To specify a voice from https://github.com/suno-ai/bark#-voice-presets ( https://suno-ai.notion.site/8b8e8749ed514b0cbf3f699013548683?v=bc67cff786b04b50b3ceb756fd05f68c ), use the `model` parameter:
+
+```
+curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{         
+     "backend": "bark",
+     "input":"Hello!",
+     "model": "v2/en_speaker_4"
+   }' | aplay
+```
+
+### Piper
+
+To install the `piper` audio models manually:
+
+- Download Voices from https://github.com/rhasspy/piper/releases/tag/v0.0.2
+- Extract the `.tar.tgz` files (.onnx,.json) inside `models`
+- Run the following command to test the model is working
+
+To use the tts endpoint, run the following command. You can specify a backend with the `backend` parameter. For example, to use the `piper` backend:
+```bash
+curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{
+  "model":"it-riccardo_fasol-x-low.onnx",
+  "backend": "piper",
+  "input": "Ciao, sono Ettore"
+}' | aplay
+```
+
+Note:
+
+- `aplay` is a Linux command. You can use other tools to play the audio file.
+- The model name is the filename with the extension.
+- The model name is case sensitive.
+- LocalAI must be compiled with the `GO_TAGS=tts` flag.
+
+### Transformers-musicgen
+
+LocalAI also has experimental support for `transformers-musicgen` for the generation of short musical compositions. Currently, this is implemented via the same requests used for text to speech:
+
+```
+curl --request POST \
+  --url http://localhost:8080/tts \
+  --header 'Content-Type: application/json' \
+  --data '{
+    "backend": "transformers-musicgen",
+    "model": "facebook/musicgen-medium",
+    "input": "Cello Rave"
+}' | aplay
+```
+
+Future versions of LocalAI will expose additional control over audio generation beyond the text prompt.
+
+### Vall-E-X
+
+[VALL-E-X](https://github.com/Plachtaa/VALL-E-X) is an open source implementation of Microsoft's VALL-E X zero-shot TTS model.
+
+#### Setup
+
+The backend will automatically download the required files in order to run the model.
+
+This is an extra backend - in the container is already available and there is nothing to do for the setup. If you are building manually, you need to install Vall-E-X manually first.
+
+#### Usage
+
+Use the tts endpoint by specifying the vall-e-x backend:
+
+```
+curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{         
+     "backend": "vall-e-x",
+     "input":"Hello!"
+   }' | aplay
+```
+
+#### Voice cloning
+
+In order to use voice cloning capabilities you must create a `YAML` configuration file to setup a model:
+
+```yaml
+name: cloned-voice
+backend: vall-e-x
+parameters:
+  model: "cloned-voice"
+vall-e:
+  # The path to the audio file to be cloned
+  # relative to the models directory 
+  audio_path: "path-to-wav-source.wav"
+```
+
+Then you can specify the model name in the requests:
+
+```
+curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{         
+     "backend": "vall-e-x",
+     "model": "cloned-voice",
+     "input":"Hello!"
+   }' | aplay
+```

docs/content/docs/getting-started/_index.en.md

@@ -0,0 +1,7 @@
+
++++
+disableToc = false
+title = "Getting started"
+weight = 2
+icon = "rocket_launch"
++++