Dockerfile

# FROM ghcr.io/ggml-org/llama.cpp:full
FROM ghcr.io/ggml-org/llama.cpp:server

WORKDIR /app

RUN apt update && apt install -y python3 python3-pip python3-venv pipx
ENV PATH=/root/.local/bin:$PATH

# RUN pip install huggingface_hub --break-system-packages
RUN pipx install huggingface_hub

# RUN python3 -c 'from huggingface_hub import hf_hub_download;  repo="prism-ml/Bonsai-8B-gguf"; hf_hub_download(repo_id=repo, filename="Bonsai-8B.gguf", local_dir="/app")'

RUN hf download prism-ml/Bonsai-8B-gguf Bonsai-8B.gguf --local-dir /app
RUN mv /app/Bonsai-8B.gguf /app/bonsai-8b

# CMD ["--server", \
CMD ["--model", "/app/bonsai-8b", \
     "--host", "0.0.0.0", \
     "--port", "7860", \
     "--threads", "2", \
     "--cache-type-k", "q8_0", \
     "--cache-type-v", "iq4_nl", \
     "-c", "65536", \
     "--log-disable", \
     "-n", "38912"]
#     "--prio", "2", \