docker-compose.yaml

version: '3.8'

services:
  train:
    build:
      context: .
    command: |
      python -m src.train_optuna_callbacks experiment=catdog_experiment ++task_name=train ++train=True ++test=False && \
      touch /app/checkpoints/train_done.flag
    volumes:
      - ./data:/app/data
      - ./checkpoints:/app/checkpoints
      - ./artifacts:/app/artifacts
      - ./logs:/app/logs
    environment:
      - PYTHONUNBUFFERED=1
      - PYTHONPATH=/app
    shm_size: '4g'
    networks:
      - default
    env_file:
      - .env
    
  eval:
    build:
      context: .
    command: |
      sh -c 'while [ ! -f /app/checkpoints/train_done.flag ]; do sleep 10; done && python -m src.train_optuna_callbacks experiment=catdog_experiment ++task_name=eval ++train=False ++test=True'
    volumes:
      - ./data:/app/data
      - ./checkpoints:/app/checkpoints
      - ./artifacts:/app/artifacts
      - ./logs:/app/logs
    environment:
      - PYTHONUNBUFFERED=1
      - PYTHONPATH=/app
    shm_size: '4g'
    networks:
      - default
    env_file:
      - .env
    depends_on:
      - train

  # Uncomment if you want to use inference service in the future
  # inference:
  #   build:
  #     context: .
  #   command: |
  #     sh -c 'while [ ! -f /app/checkpoints/train_done.flag ]; do sleep 10; done && python -m src.infer experiment=catdog_experiment'
  #   volumes:
  #     - ./data:/app/data
  #     - ./checkpoints:/app/checkpoints
  #     - ./artifacts:/app/artifacts
  #     - ./logs:/app/logs
  #   environment:
  #     - PYTHONUNBUFFERED=1
  #     - PYTHONPATH=/app
  #   shm_size: '4g'
  #   networks:
  #     - default
  #   env_file:
  #     - .env  
  #   depends_on:
  #     - train

volumes:
  data:
  checkpoints:
  artifacts:
  logs:

networks:
  default: