services:
  ollama:
    image: ollama/ollama:latest
    container_name: ollama
    restart: unless-stopped
    ports:
      - "11434:11434"
    volumes:
      - /mnt/e/volumes/ollama/data:/root/.ollama
    environment:
      - OLLAMA_HOST=0.0.0.0:11434
      # Optional: Set GPU device if you have multiple GPUs
      # - NVIDIA_VISIBLE_DEVICES=0
    command: serve
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    networks:
      - app-network
    healthcheck:
      test: ["CMD", "ollama", "list"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s

  faster-whisper:
    image: lscr.io/linuxserver/faster-whisper:gpu-legacy
    container_name: faster-whisper
    gpus: all
    environment:
      - PUID=1000
      - PGID=1000
      - TZ=Europe/Paris
      - WHISPER_MODEL=turbo    # bon compromis pour RTX 4060 Ti
      - WHISPER_LANG=fr
      - WHISPER_BEAM=5         # précision vs rapidité
    volumes:
      - /mnt/e/volumes/faster-whisper/audio:/app
      - /mnt/e/volumes/faster-whisper/models:/root/.cache/whisper
    ports:
      - 10300:10300
    restart: unless-stopped

networks:
  app-network:
    driver: bridge