services:
  vllm-qwen:
    build: 
      context: .
      dockerfile: Dockerfile
    container_name: vllm_qwen_08b
    ports:
      - "8000:8000"
    ipc: host 
    devices:
      - "/dev/kfd:/dev/kfd"
      - "/dev/dri:/dev/dri"
    group_add:
      - video
      - render
    volumes:
      # Montaggio in sola lettura del modello
      - /opt/models:/app/models:ro
    environment:
      - HSA_OVERRIDE_GFX_VERSION=11.0.0
      - HIP_VISIBLE_DEVICES=0
    command: >
      --model /app/models/Qwen3.5-0.8B-UD-Q8_K_XL.gguf
      --quantization gguf
      --gpu-memory-utilization 0.50
      --max-model-len 4096
    restart: unless-stopped