vLLM Project Update

Update from the maintainers of vLLM, the high-throughput and memory-efficient inference and serving engine for LLMs.