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

WORKDIR /app

RUN apt update && apt install -y python3 python3-pip python3-venv
RUN python3 -m venv /opt/venv
ENV PATH="/opt/venv/bin:$PATH"

RUN pip install -U pip huggingface_hub

# Download BOTH the main AI and the dspark speedup layer
RUN python3 -c 'from huggingface_hub import hf_hub_download; \
    repo="prism-ml/Ternary-Bonsai-27B-gguf"; \
    hf_hub_download(repo_id=repo, filename="Ternary-Bonsai-27B-Q2_0.gguf", local_dir="/models"); \
    hf_hub_download(repo_id=repo, filename="Ternary-Bonsai-27B-dspark-Q4_1.gguf", local_dir="/models")'

# Run the main model (-m) alongside the draft model (-md)
CMD ["--server", \
     "-m", "/models/Ternary-Bonsai-27B-Q2_0.gguf", \
     "-md", "/models/Ternary-Bonsai-27B-dspark-Q4_1.gguf", \
     "--draft", "4", \
     "--host", "0.0.0.0", \
     "--port", "7860", \
     "-t", "2", \
     "--cache-type-k", "f16", \
     "--cache-type-v", "f16", \
     "-c", "8024", \
     "-n", "4096"]