{"id":"vllm","name":"vLLM","desc":"基于PagedAttention的高吞吐LLM推理引擎，显著提升GPU利用率","url":"https://github.com/vllm-project/vllm","category":"开源项目","tags":["推理","高性能","GPU"],"rating":0,"users":null,"updated":"2026-04-08T00:00:00.000Z","source":"roadmap.sh/ai-engineer","addedAt":"2026-04-08T00:00:00.000Z","hot":false,"isNew":false}