Add vLLM as one of the supported inference engines.

Create README.md

Upload checkpoint

initial commit