Dockerfile.llama.cpu

# Dockerfile.llama.cpu
FROM ubuntu:24.04

ENV DEBIAN_FRONTEND=noninteractive

# Install build dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \
    build-essential git cmake curl libcurl4-openssl-dev libssl-dev ca-certificates libgomp1 \
    && rm -rf /var/lib/apt/lists/*

# Clone and build llama.cpp from source
RUN git clone https://github.com/ggerganov/llama.cpp.git /llama.cpp
WORKDIR /llama.cpp

# Use cmake for building
RUN cmake -B build -DLLAMA_CURL=ON -DLLAMA_OPENSSL=ON -DGGML_CUDA=OFF
RUN cmake --build build --config Release -j $(nproc)

# Create a non-root user and models directory
RUN useradd -m -s /bin/bash llamauser \
    && mkdir -p /models && chown -R llamauser:llamauser /models /llama.cpp

# Switch to the non-root user
USER llamauser

# Set working directory to the bin folder and ensure libraries are found
WORKDIR /llama.cpp/build/bin
ENV LD_LIBRARY_PATH=/llama.cpp/build/bin:$LD_LIBRARY_PATH

ENTRYPOINT ["./llama-server"]