FROM nvidia/cuda:11.8.0-devel-ubuntu22.04
# Install necessary packages
RUN apt update && apt install git build-essential libopenblas-dev wget python3-pip -y
RUN apt update && apt install -y wget
RUN apt-get install -y nodejs npm
RUN npm install -g localtunnel
# Set up a new user named "user" with user ID 1000
RUN useradd -m -u 1000 user
# Switch to the "user" user
USER user
# Set home to the user's home directory
ENV HOME=/home/user \
    PATH=/home/user/.local/bin:$PATH
# Set the working directory to the user's home directory
WORKDIR $HOME/app
# Create a virtual environment and activate it
RUN python3 -m venv venv
ENV PATH="$HOME/app/venv/bin:$PATH"
# Copy the current directory contents into the container at $HOME/app setting the owner to the user
COPY --chown=user . $HOME/app
# Install the package in editable mode
RUN git clone https://github.com/lolcats399/aphrodite-engine && cd aphrodite-engine && pip3 install -e .
RUN pip install huggingface-hub hf-transfer
ENV HF_HUB_ENABLE_HF_TRANSFER=1
RUN huggingface-cli download TheBloke/goliath-120b-gptq --local-dir ~/goliath-gptq --local-dir-use-symlinks False --cache-dir ~/cache
RUN echo "#!/bin/bash" > startup.sh
RUN echo "lt --port 7860 &" >> startup.sh
RUN echo "public_ip=\$(wget -qO- https://loca.lt/mytunnelpassword)" >> startup.sh
RUN echo "echo \"Public IP: \$public_ip\"" >> startup.sh
RUN chmod +x startup.sh
CMD /bin/bash -c "./startup.sh && python3 $HOME/app/aphrodite-engine/aphrodite/endpoints/openai/api_server.py $ENGINE_ARGS --port 7860 --model ~/goliath-gptq"