services:
  ai-server:
    build:
      context: .
      dockerfile: ./Dockerfile
    runtime: nvidia
    volumes:
      - models:/models/
      - /root/.cache/huggingface:/root/.cache/huggingface
    environment:
      - MODEL_LIBRARY=/models/
      - NVIDIA_VISIBLE_DEVICES=all
    ports:
      - "8000:8000"

volumes:
  models: