New top story on Hacker News: Consistency LLM: converting LLMs to parallel decoders accelerates inference 3.5x

No comments