Dockerfile

FROM debian:bookworm-slim

ARG DEBIAN_FRONTEND=noninteractive
RUN apt-get update && apt-get install -y --no-install-recommends \
    git build-essential cmake curl ca-certificates pkg-config libcurl4-openssl-dev \
 && rm -rf /var/lib/apt/lists/*

# build llama.cpp (HTTP server in tools/server)
WORKDIR /app
RUN git clone --depth 1 https://github.com/ggml-org/llama.cpp.git \
 && cd llama.cpp && mkdir -p build && cd build \
 && cmake -DCMAKE_BUILD_TYPE=Release \
          -DGGML_NATIVE=ON \
          -DLLAMA_BUILD_EXAMPLES=ON \
          -DLLAMA_BUILD_SERVER=ON \
          -DLLAMA_BUILD_TESTS=OFF \
          .. \
 && cmake --build . --target llama-server -j

# fetch LiquidAI LFM2-350M GGUF
RUN mkdir -p /models && \
    curl -fL --retry 5 --retry-delay 2 -o /models/model.gguf \
    "https://huggingface.co/LiquidAI/LFM2-350M-GGUF/resolve/main/LFM2-350M-Q4_K_M.gguf?download=true"

EXPOSE 7860
CMD ["/app/llama.cpp/build/bin/llama-server","-m","/models/model.gguf","-c","2048","-ngl","0","-t","4","--host","0.0.0.0","--port","7860"]