---

docker run --gpus all \
  -v ~/vllm/models/qwen/:/models/qwen \
  -p 8000:8000 \
  --ipc=host \
  vllm/vllm-openai:latest \
  --model /models/qwen \
  --tensor-parallel-size 2