services:
  docling_app:
    build:
      context: .
      args:
        ENABLE_CUDA: '1'
    container_name: docling_app
    restart: unless-stopped
    ports:
      - "7880:7860"  # Gradio application port
    volumes:
      - ./app/services/weaviate_service.py:/gradio_app/weaviate_service.py
    environment:
      ENABLE_CUDA: '1'
      NVIDIA_VISIBLE_DEVICES: '2'
    deploy:
        resources:
            reservations:
              devices:
                - driver: nvidia
                  device_ids: ['2']  # Specify GPU 3
                  capabilities: [gpu]
    shm_size: 1g