# Base image: Hugging Face TGI
FROM ghcr.io/huggingface/text-generation-inference:3.0.2

# Set working directory
WORKDIR /app

# Minimal storage directories
RUN mkdir -p /data && chmod 777 /data
RUN mkdir -p /.cache && chmod 777 /.cache
RUN mkdir -p /.triton && chmod 777 /.triton

# Expose the API port
EXPOSE 8080

# Optional: Set Hugging Face token for private model access
ARG HF_TOKEN
ENV HF_TOKEN=${HF_TOKEN}

# Model: Replace with any small or quantized TGI-compatible model
# Example: TheBloke/Mistral-7B-Instruct-v0.1-GPTQ or TinyLlama/TinyLlama-1.1B-Chat
CMD ["--model-id", "TinyLlama/TinyLlama-1.1B-Chat-v1.0", "--port", "8080"]