Dockerfile-cuda

# Start from an NVIDIA CUDA base image with Python 3
FROM nvidia/cuda:11.8.0-cudnn8-runtime-ubuntu22.04

LABEL org.opencontainers.image.source=https://github.com/protectai/llm-guard
LABEL org.opencontainers.image.description="LLM Guard API"
LABEL org.opencontainers.image.licenses=MIT

# Install Python and other necessary packages
RUN apt-get update && apt-get install -y \
    python3-pip \
    python3-dev \
    build-essential \
    && apt-get clean && rm -rf /var/lib/apt/lists/*

# Alias python3 to python
RUN ln -s /usr/bin/python3 /usr/bin/python

# Create a non-root user and set user environment variables
RUN useradd -m -u 1000 user
USER user
ENV HOME=/home/user \
    PATH=/home/user/.local/bin:$PATH

# ensures that the python output is sent straight to terminal (e.g. your container log)
# without being first buffered and that you can see the output of your application (e.g. django logs)
# in real time. Equivalent to python -u: https://docs.python.org/3/using/cmdline.html#cmdoption-u
ENV PYTHONUNBUFFERED 1

# https://docs.python.org/3/using/cmdline.html#envvar-PYTHONDONTWRITEBYTECODE
# Prevents Python from writing .pyc files to disk
ENV PYTHONDONTWRITEBYTECODE 1

# Set up a working directory
WORKDIR $HOME/app

# Copy pyproject.toml and other necessary files for installation
COPY --chown=user:user pyproject.toml ./
COPY --chown=user:user app ./app

# Install the project's dependencies
RUN pip3 install --no-cache-dir --upgrade pip && \
    pip3 install --no-cache-dir torch==2.0.1 --extra-index-url https://download.pytorch.org/whl/cu118 && \
    pip3 install --no-cache-dir ".[gpu]"

RUN python -m spacy download en_core_web_sm

COPY --chown=user:user ./config/scanners.yml ./config/scanners.yml

# Expose the port the app runs on
EXPOSE 7860

# Specify the default command
CMD ["llm_guard_api", "/home/user/app/config/scanners.yml"]