Dockerfile

FROM python:3.11-slim

RUN apt-get update && \
    apt-get install -y libopenblas-dev ninja-build build-essential pkg-config curl

RUN pip install -U pip setuptools wheel && \
    CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS" FORCE_CMAKE=1 pip install --verbose llama-cpp-python[server]

RUN mkdir models && \
    curl -L https://huggingface.co/TheBloke/openchat_3.5-GGUF/resolve/main/openchat_3.5.Q4_K_M.gguf -o models/gguf-model.bin

COPY ./startserver.sh ./
COPY ./main.py ./
COPY ./index.html ./

RUN chmod +x ./startserver.sh

ENV HOST=0.0.0.0
ENV PORT=7860
EXPOSE ${PORT}

# start server 
CMD ["/bin/sh", "./startserver.sh"]