Dockerfile

FROM ghcr.io/ggerganov/llama.cpp:full-cuda

ENV DEBIAN_FRONTEND=noninteractive

# Update and install necessary dependencies
RUN apt update && \
    apt install --no-install-recommends -y \
        build-essential \
        python3 \
        python3-pip \
        wget \
        curl \
        git \
        cmake \
        zlib1g-dev \
        libblas-dev && \
    apt clean && \
    rm -rf /var/lib/apt/lists/*

# Setting up CUDA environment variables (this may not be necessary since you're using the official nvidia/cuda image, but it's good to be explicit)
ENV PATH="/usr/local/cuda/bin:$PATH" \
    LD_LIBRARY_PATH="/usr/local/cuda/lib64:$LD_LIBRARY_PATH" \
    CUDA_HOME="/usr/local/cuda"

WORKDIR /app

RUN wget https://huggingface.co/itsdotscience/Magicoder-S-DS-6.7B-GGUF/resolve/main/Magicoder-S-DS-6.7B_q8_0.gguf

RUN make LLAMA_CUBLAS=1

# Expose the port
EXPOSE 8080

# Start the llava-server with models
CMD ["--server", "--model", "Magicoder-S-DS-6.7B_q8_0.gguf", "--threads", "6", "--host", "0.0.0.0", "-ngl", "33"]