services:
  inference:
    build:
      context: .
      dockerfile: Dockerfile
    image: rocm-vllm-inference:latest
    container_name: rocm-vllm-inference
    command: ["python", "-m", "app.start_api"]
    ports:
      - "8000:8000"
    volumes:
      - /opt/model:/opt/model:ro
      - ./config.json:/workspace/config.json:ro
    devices:
      - /dev/kfd
      - /dev/dri
    group_add:
      - video
    ipc: host
    shm_size: 16g
    cap_add:
      - SYS_PTRACE
    security_opt:
      - seccomp=unconfined
    restart: unless-stopped
  openai_api:
    build:
      context: .
      dockerfile: Dockerfile
    image: rocm-vllm-inference:latest
    container_name: rocm-vllm-openai
    command: ["python", "-m", "app.start_openai"]
    ports:
      - "8001:8001"
    volumes:
      - /opt/model:/opt/model:ro
      - ./config.json:/workspace/config.json:ro
    devices:
      - /dev/kfd
      - /dev/dri
    group_add:
      - video
    ipc: host
    shm_size: 16g
    cap_add:
      - SYS_PTRACE
    security_opt:
      - seccomp=unconfined
    restart: unless-stopped