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

RUN apt update && apt install wget -y && rm -rf /var/lib/apt/lists/*

# 1. THE MAIN MODEL: Using Q4_K_M for speed on 2vCPUs
RUN wget "https://huggingface.co/unsloth/gemma-3n-E2B-it-GGUF/resolve/main/gemma-3n-E2B-it-Q4_K_M.gguf" -O /model.gguf

# Run the server natively (Text-Only Mode)
# -c 4096 is the safety limit to prevent RAM crashes
ENTRYPOINT []
CMD /app/llama-server -m /model.gguf -t 2 --port 7860 --host 0.0.0.0 -c 4096 --temp 1.0 --log-disable > /dev/null 2>&1