Setup CICD to build and push images to Dockerhub

.github/workflows/main.yaml

@@ -2,23 +2,104 @@ on:
   push:
     branches:
       - main
+  pull_request:
   workflow_dispatch:
 
 jobs:
-  build-root-gpu-image:
+  # builds all the base images required to build the full GPU-capable CUDA-enabled Docker images 
+  build-base-gpu-image:
     runs-on: ubuntu-latest
     strategy:
+      fail-fast: true
       matrix:
-        cuda: [12.0.1, 12.1.1, 12.2.2, 12.3.2, 12.4.1, 12.5.0]
+        cuda: [12.5.0] # [12.0.1, 12.1.1, 12.2.2, 12.3.2, 12.4.1, 12.5.0]
     steps:
       - name: 'Checkout GitHub Action'
         uses: actions/checkout@main
+      - name: 'Login to GitHub Container Registry'
+        uses: docker/login-action@v1
+        with:
+          registry: ghcr.io
+          username: ${{github.actor}}
+          password: ${{secrets.GITHUB_TOKEN}}
+      - name: 'Build Base GPU Image'
+        run: CUDA_VERSION=${{matrix.cuda}} ./ci/bash/build_base_gpu_image.sh
+  # builds the base image required to build the ONX CPU-bound Docker images
+  build-base-cpu-image:
+    runs-on: ubuntu-latest
+    steps:
+      - name: 'Checkout GitHub Action'
+        uses: actions/checkout@main
+      - name: 'Login to GitHub Container Registry'
+        uses: docker/login-action@v1
+        with:
+          registry: ghcr.io
+          username: ${{github.actor}}
+          password: ${{secrets.GITHUB_TOKEN}}
+      - name: 'Build Base CPU Image'
+        run: ./ci/bash/build_base_cpu_image.sh
+
+  # builds the ONNX images for the available models
+  build-cpu-image:
+    runs-on: ubuntu-latest
+    needs: [build-base-cpu-image]
+    strategy:
+      fail-fast: false
+      matrix:
+        model: [
+          BAAI/bge-small-en-v1.5,
+          # BAAI/bge-medium-en-v1.5,
+          # BAAI/bge-large-en-v1.5,
+          # Snowflake/snowflake-artic-embed-xs,
+          # Snowflake/snowflake-artic-embed-s,
+          # Snowflake/snowflake-artic-embed-m,
+          # Snowflake/snowflake-artic-embed-l,
+          # mixedbread-ai/mxbai-embed-large-v1,
+          # sentence-transformers/all-MiniLM-L6-v2,
+          # sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
+        ]
+        targetarch: [arm64] #[amd64, arm64]
+    steps:
+      - name: 'Checkout GitHub Action'
+        uses: actions/checkout@main
+      - name: 'Login to GitHub Container Registry'
+        uses: docker/login-action@v1
+        with:
+          registry: ghcr.io
+          username: ${{github.actor}}
+          password: ${{secrets.GITHUB_TOKEN}}
+      - name: 'Build Production CPU Image'
+        run: docker build . --build-arg MODEL_NAME=${{matrix.model}} --build-arg TARGETARCH=${{matrix.targetarch}} --file Dockerfile.cpu --target release
 
+  # builds the CUDA images for the available models
+  build-gpu-image:
+    runs-on: gpu-t4-4-core
+    needs: [build-base-gpu-image]
+    strategy:
+      fail-fast: false
+      matrix:
+        cuda: [12.5.0] # [12.0.1, 12.1.1, 12.2.2, 12.3.2, 12.4.1, 12.5.0]
+        model: [
+          BAAI/bge-small-en-v1.5,
+          # BAAI/bge-medium-en-v1.5,
+          # BAAI/bge-large-en-v1.5,
+          # Snowflake/snowflake-artic-embed-xs,
+          # Snowflake/snowflake-artic-embed-s,
+          # Snowflake/snowflake-artic-embed-m,
+          # Snowflake/snowflake-artic-embed-l,
+          # mixedbread-ai/mxbai-embed-large-v1,
+          # sentence-transformers/all-MiniLM-L6-v2,
+          # sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
+        ]
+        targetarch: [arm64] #[amd64, arm64]
+    steps:
+      - name: 'Checkout GitHub Action'
+        uses: actions/checkout@main
       - name: 'Login to GitHub Container Registry'
         uses: docker/login-action@v1
         with:
           registry: ghcr.io
           username: ${{github.actor}}
           password: ${{secrets.GITHUB_TOKEN}}
-      - name: 'Build Root GPU Image'
-        run: CUDA_VERSION=${{matrix.cuda}} ./ci/bash/build_root_gpu_image.sh
+      - name: 'Build Production GPU Image'
+        run: docker build . --build-arg MODEL_NAME=${{matrix.model}} --build-arg TARGETARCH=${{matrix.targetarch}} --build-arg CUDA_VERSION=${{matrix.cuda}} --file Dockerfile.gpu --target release

Dockerfile.root.cpu → Dockerfile.base.cpu

@@ -1,4 +1,5 @@
 FROM ubuntu:22.04
+LABEL org.opencontainers.image.source https://github.com/weaviate/t2v-transformers-models-rs
 
 # Install rust dependencies
 RUN apt-get update -qq \

Dockerfile.cpu

@@ -1,35 +1,12 @@
-FROM ubuntu:22.04 as sys
-
-# Install rust dependencies
-RUN apt-get update -qq \
-    && apt-get install -qq -y vim gcc g++ curl git build-essential libssl-dev openssl pkg-config \
-    && rm -rf /var/lib/apt/lists/*
-
-# Install Rust
-RUN curl https://sh.rustup.rs -sSf | bash -s -- -y
-ENV PATH="/root/.cargo/bin:${PATH}"
-
-# Install Python
-RUN apt-get update && \
-    apt-get install -y python3-pip python3-dev python3-venv && \
-    rm -rf /var/lib/apt/lists/*
-
+FROM ghcr.io/weaviate/t2v-transformers-models-rs/base-cpu:latest as build
+RUN . venv/bin/activate
 WORKDIR /app
 
 # Copy our manifests
-COPY ./Cargo.lock ./Cargo.lock
 COPY ./Cargo.toml ./Cargo.toml
 COPY ./src ./src
 COPY ./ci ./ci
 
-FROM sys as build
-WORKDIR /app
-
-RUN python3 -m venv venv
-RUN . venv/bin/activate
-# Install Python scripts for model download
-RUN pip install -r ci/requirements.txt
-
 ARG TARGETARCH
 ARG MODEL_NAME
 

Dockerfile.gpu

@@ -1,11 +1,10 @@
 ARG CUDA_VERSION
-FROM ghcr.io/weaviate/t2v-transformers-models-rs/t2v-rs-root-gpu-${CUDA_VERSION}:0.0.1 as build
+FROM ghcr.io/weaviate/t2v-transformers-models-rs/base-gpu-${CUDA_VERSION}:latest as build
 RUN . venv/bin/activate
 
 WORKDIR /app
 
 # Copy our manifests
-COPY ./Cargo.lock ./Cargo.lock
 COPY ./Cargo.toml ./Cargo.toml
 COPY ./src ./src
 COPY ./ci ./ci

ci/bash/build_base_cpu_image.sh

@@ -0,0 +1,10 @@
+DOCKER_REGISTRY="ghcr.io/weaviate/t2v-transformers-models-rs"
+IMAGE=$DOCKER_REGISTRY/base-cpu:latest
+
+docker build --cache-from $IMAGE -t base-cpu -f Dockerfile.base.cpu .
+
+docker tag base-cpu $IMAGE
+
+docker push $IMAGE
+
+docker logout

ci/bash/build_base_gpu_image.sh

@@ -0,0 +1,10 @@
+DOCKER_REGISTRY="ghcr.io/weaviate/t2v-transformers-models-rs"
+IMAGE=$DOCKER_REGISTRY/base-gpu-${CUDA_VERSION}:latest
+
+docker build --cache-from $IMAGE --build-arg CUDA_VERSION=$CUDA_VERSION -t base-gpu -f Dockerfile.base.gpu .
+
+docker tag base-gpu $IMAGE
+
+docker push $IMAGE
+
+docker logout

docker-compose.yml

@@ -1,50 +1,50 @@
 ---
 version: '3.4'
 services:
-  t2v-transformers-rs-gpu:
-    build:
-      context: .
-      dockerfile: Dockerfile.gpu
-      target: release
-      args:
-        - MODEL_NAME: "BAAI/bge-small-en-v1.5"
-        - TARGETARCH: "arm64"
-        - CUDA_VERSION: "12.4.1"
-    ports:
-      - 3000:3000
-    deploy:
-      resources:
-        reservations:
-          devices:
-            - driver: nvidia
-              count: 1
-              capabilities: [gpu]
-  t2v-transformers-rs-cpu:
-    build:
-      context: .
-      dockerfile: Dockerfile.cpu
-      target: release
-      args:
-        - MODEL_NAME: "BAAI/bge-small-en-v1.5"
-        - TARGETARCH: "arm64"
-    ports:
-      - 3000:3000
-  t2v-transformers-py-gpu:
-    image: semitechnologies/transformers-inference:baai-bge-small-en-v1.5
-    ports:
-      - 8080:8080
-    environment:
-      ENABLE_CUDA: "true"
-      T2V_TRANSFORMERS_DIRECT_TOKENIZE: "true"
-    deploy:
-      resources:
-        reservations:
-          devices:
-            - driver: nvidia
-              count: 1
-              capabilities: [gpu]
+  # t2v-transformers-rs-gpu:
+  #   build:
+  #     context: .
+  #     dockerfile: Dockerfile.gpu
+  #     target: release
+  #     args:
+  #       - MODEL_NAME: "BAAI/bge-small-en-v1.5"
+  #       - TARGETARCH: "arm64"
+  #       - CUDA_VERSION: "12.4.1"
+  #   ports:
+  #     - 3000:3000
+  #   deploy:
+  #     resources:
+  #       reservations:
+  #         devices:
+  #           - driver: nvidia
+  #             count: 1
+  #             capabilities: [gpu]
+  # t2v-transformers-rs-cpu:
+  #   build:
+  #     context: .
+  #     dockerfile: Dockerfile.cpu
+  #     target: release
+  #     args:
+  #       - MODEL_NAME: "BAAI/bge-small-en-v1.5"
+  #       - TARGETARCH: "arm64"
+  #   ports:
+  #     - 3000:3000
+  # t2v-transformers-py-gpu:
+  #   image: semitechnologies/transformers-inference:baai-bge-small-en-v1.5
+  #   ports:
+  #     - 8080:8080
+  #   environment:
+  #     ENABLE_CUDA: "true"
+  #     T2V_TRANSFORMERS_DIRECT_TOKENIZE: "true"
+  #   deploy:
+  #     resources:
+  #       reservations:
+  #         devices:
+  #           - driver: nvidia
+  #             count: 1
+  #             capabilities: [gpu]
   t2v-transformers-py-cpu:
-    image: semitechnologies/transformers-inference:baai-bge-small-en-v1.5-onnx
+    image: semitechnologies/transformers-inference:mixedbread-ai-mxbai-embed-large-v1-onnx
     ports:
       - 8080:8080
     environment: