FROM ollama/ollama:latest

# Provision network tooling assets
RUN apt-get update && apt-get install -y curl wget && rm -rf /var/lib/apt/lists/*

WORKDIR /app

# Pull the lightweight Q4_K_M GGUF format mapping from bartowski's mirror
RUN wget https://huggingface.co/bartowski/WhiteRabbitNeo-2.5-Qwen-2.5-Coder-7B-GGUF/resolve/main/WhiteRabbitNeo-2.5-Qwen-2.5-Coder-7B-Q4_K_M.gguf -O model.gguf

# Write constraints config (Force 2 threads to prevent free CPU workspace context freezes)
# RUN printf "FROM ./model.gguf\nPARAMETER num_ctx 4096\nPARAMETER num_thread 2" > Modelfile
RUN printf "FROM ./model.gguf\nPARAMETER num_ctx 2048\nPARAMETER num_thread 4" > Modelfile

# Open operational network port routes
EXPOSE 7860
ENV OLLAMA_HOST=0.0.0.0:7860
ENV OLLAMA_KEEP_ALIVE=-1

COPY entrypoint.sh /entrypoint.sh
RUN chmod +x /entrypoint.sh

ENTRYPOINT ["/entrypoint.sh"]