[core / Docker] Add workflow to build TRL docker images (#1215)

* add docker build * Update docker/trl-latest-gpu/Dockerfile * Update docker/trl-source-gpu/Dockerfile
huggingface · Jan 11, 2024 · 2eeb7b0 · 2eeb7b0
1 parent 9f8d0e4
commit 2eeb7b0
Show file tree

Hide file tree

Showing 3 changed files with 206 additions and 0 deletions.
diff --git a/.github/workflows/docker-build.yml b/.github/workflows/docker-build.yml
@@ -0,0 +1,74 @@
+name: Build Docker images (scheduled)
+
+on:
+  workflow_dispatch:
+  workflow_call:
+  schedule:
+    - cron: "0 1 * * *"
+
+concurrency:
+  group: docker-image-builds
+  cancel-in-progress: false
+
+jobs:
+  trl-latest:
+    name: "Latest TRL GPU"
+    runs-on: ubuntu-latest
+    steps:
+      - name: Cleanup disk
+        run: |
+          sudo ls -l /usr/local/lib/
+          sudo ls -l /usr/share/
+          sudo du -sh /usr/local/lib/
+          sudo du -sh /usr/share/
+          sudo rm -rf /usr/local/lib/android
+          sudo rm -rf /usr/share/dotnet
+          sudo du -sh /usr/local/lib/
+          sudo du -sh /usr/share/
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v1
+      - name: Check out code
+        uses: actions/checkout@v3
+      - name: Login to DockerHub
+        uses: docker/login-action@v1
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_PASSWORD }}
+
+      - name: Build and Push GPU
+        uses: docker/build-push-action@v4
+        with:
+          context: ./docker/trl-latest-gpu
+          push: true
+          tags: huggingface/trl-latest-gpu
+
+  trl-source:
+    name: "Latest TRL + HF ecosystem from source"
+    runs-on: ubuntu-latest
+    steps:
+      - name: Cleanup disk
+        run: |
+          sudo ls -l /usr/local/lib/
+          sudo ls -l /usr/share/
+          sudo du -sh /usr/local/lib/
+          sudo du -sh /usr/share/
+          sudo rm -rf /usr/local/lib/android
+          sudo rm -rf /usr/share/dotnet
+          sudo du -sh /usr/local/lib/
+          sudo du -sh /usr/share/
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v1
+      - name: Check out code
+        uses: actions/checkout@v3
+      - name: Login to DockerHub
+        uses: docker/login-action@v1
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_PASSWORD }}
+
+      - name: Build and Push GPU
+        uses: docker/build-push-action@v4
+        with:
+          context: ./docker/trl-source-gpu
+          push: true
+          tags: huggingface/trl-source-gpu
diff --git a/docker/trl-latest-gpu/Dockerfile b/docker/trl-latest-gpu/Dockerfile
@@ -0,0 +1,66 @@
+# Builds GPU docker image of PyTorch
+# Uses multi-staged approach to reduce size
+# Stage 1
+# Use base conda image to reduce time
+FROM continuumio/miniconda3:latest AS compile-image
+# Specify py version
+ENV PYTHON_VERSION=3.10
+# Install apt libs - copied from https://github.com/huggingface/accelerate/blob/main/docker/accelerate-gpu/Dockerfile
+RUN apt-get update && \
+    apt-get install -y curl git wget software-properties-common git-lfs && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists*
+
+# Install audio-related libraries 
+RUN apt-get update && \
+    apt install -y ffmpeg
+
+RUN apt install -y libsndfile1-dev
+RUN git lfs install
+
+# Create our conda env - copied from https://github.com/huggingface/accelerate/blob/main/docker/accelerate-gpu/Dockerfile
+RUN conda create --name trl python=${PYTHON_VERSION} ipython jupyter pip
+RUN python3 -m pip install --no-cache-dir --upgrade pip
+
+# Below is copied from https://github.com/huggingface/accelerate/blob/main/docker/accelerate-gpu/Dockerfile
+# We don't install pytorch here yet since CUDA isn't available
+# instead we use the direct torch wheel
+ENV PATH /opt/conda/envs/trl/bin:$PATH
+# Activate our bash shell
+RUN chsh -s /bin/bash
+SHELL ["/bin/bash", "-c"]
+
+# Stage 2
+FROM nvidia/cuda:12.2.2-devel-ubuntu22.04 AS build-image
+COPY --from=compile-image /opt/conda /opt/conda
+ENV PATH /opt/conda/bin:$PATH
+
+RUN chsh -s /bin/bash
+SHELL ["/bin/bash", "-c"]
+RUN source activate trl && \ 
+    python3 -m pip install --no-cache-dir bitsandbytes optimum auto-gptq
+
+# Install apt libs
+RUN apt-get update && \
+    apt-get install -y curl git wget && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists*
+
+# Activate the conda env and install transformers + accelerate from source
+RUN source activate trl && \
+    python3 -m pip install -U --no-cache-dir \
+    librosa \
+    "soundfile>=0.12.1" \
+    scipy \
+    transformers \
+    accelerate \
+    peft \
+    trl
+
+RUN source activate trl && \ 
+    pip freeze | grep trl
+
+RUN echo "source activate trl" >> ~/.profile
+
+# Activate the virtualenv
+CMD ["/bin/bash"]
diff --git a/docker/trl-source-gpu/Dockerfile b/docker/trl-source-gpu/Dockerfile
@@ -0,0 +1,66 @@
+# Builds GPU docker image of PyTorch
+# Uses multi-staged approach to reduce size
+# Stage 1
+# Use base conda image to reduce time
+FROM continuumio/miniconda3:latest AS compile-image
+# Specify py version
+ENV PYTHON_VERSION=3.10
+# Install apt libs - copied from https://github.com/huggingface/accelerate/blob/main/docker/accelerate-gpu/Dockerfile
+RUN apt-get update && \
+    apt-get install -y curl git wget software-properties-common git-lfs && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists*
+
+# Install audio-related libraries 
+RUN apt-get update && \
+    apt install -y ffmpeg
+
+RUN apt install -y libsndfile1-dev
+RUN git lfs install
+
+# Create our conda env - copied from https://github.com/huggingface/accelerate/blob/main/docker/accelerate-gpu/Dockerfile
+RUN conda create --name trl python=${PYTHON_VERSION} ipython jupyter pip
+RUN python3 -m pip install --no-cache-dir --upgrade pip
+
+# Below is copied from https://github.com/huggingface/accelerate/blob/main/docker/accelerate-gpu/Dockerfile
+# We don't install pytorch here yet since CUDA isn't available
+# instead we use the direct torch wheel
+ENV PATH /opt/conda/envs/trl/bin:$PATH
+# Activate our bash shell
+RUN chsh -s /bin/bash
+SHELL ["/bin/bash", "-c"]
+
+# Stage 2
+FROM nvidia/cuda:12.2.2-devel-ubuntu22.04 AS build-image
+COPY --from=compile-image /opt/conda /opt/conda
+ENV PATH /opt/conda/bin:$PATH
+
+RUN chsh -s /bin/bash
+SHELL ["/bin/bash", "-c"]
+RUN source activate trl && \ 
+    python3 -m pip install --no-cache-dir bitsandbytes optimum auto-gptq
+
+# Install apt libs
+RUN apt-get update && \
+    apt-get install -y curl git wget && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists*
+
+# Activate the conda env and install transformers + accelerate from source
+RUN source activate trl && \
+    python3 -m pip install -U --no-cache-dir \
+    librosa \
+    "soundfile>=0.12.1" \
+    scipy \
+    git+https://github.com/huggingface/transformers \
+    git+https://github.com/huggingface/accelerate \
+    git+https://github.com/huggingface/peft \
+    trl[test]@git+https://github.com/huggingface/trl
+
+RUN source activate trl && \ 
+    pip freeze | grep transformers
+
+RUN echo "source activate trl" >> ~/.profile
+
+# Activate the virtualenv
+CMD ["/bin/bash"]