M.I.M.I.R - Multi-agent Intelligent Memory & Insight Repository

# Build llama.cpp server for AMD64 (Windows/Linux x86_64) with Vision Models # Supports both Llama 3.2 Vision (2B/11B) and Qwen2.5-VL (2B/7B) for image understanding FROM ubuntu:22.04 AS builder # Install build dependencies RUN apt-get update && apt-get install -y \ build-essential \ cmake \ git \ curl \ libcurl4-openssl-dev \ && rm -rf /var/lib/apt/lists/* # Clone llama.cpp repository WORKDIR /build RUN git clone https://github.com/ggerganov/llama.cpp.git . # Build llama.cpp server with multimodal support RUN mkdir build && cd build && \ cmake .. \ -DLLAMA_CURL=ON \ -DLLAMA_BUILD_SERVER=ON \ -DBUILD_SHARED_LIBS=OFF \ -DCMAKE_BUILD_TYPE=Release && \ cmake --build . --config Release -j$(nproc) # Runtime stage FROM ubuntu:22.04 # Install runtime dependencies RUN apt-get update && apt-get install -y \ libcurl4 \ curl \ libgomp1 \ && rm -rf /var/lib/apt/lists/* # Copy built binaries COPY --from=builder /build/build/bin/llama-server /usr/local/bin/llama-server # Create models directory RUN mkdir -p /models # Copy vision model (single model per image) ARG MODEL_FILE ARG MMPROJ_FILE COPY docker/llama-cpp/models/${MODEL_FILE} /models/model.gguf COPY docker/llama-cpp/models/${MMPROJ_FILE} /models/mmproj.gguf # Expose port EXPOSE 8080 # Health check HEALTHCHECK --interval=30s --timeout=10s --start-period=30s --retries=3 \ CMD curl -f http://localhost:8080/health || exit 1 # Set working directory WORKDIR /app # Run llama-server with vision capabilities # --mmproj is required for vision understanding ENTRYPOINT ["/usr/local/bin/llama-server"] CMD ["--host", "0.0.0.0", "--port", "8080", "--model", "/models/model.gguf", "--mmproj", "/models/mmproj.gguf", "--ctx-size", "32768", "--parallel", "4", "--threads", "-1"]