Google's TurboQuant AI-compression algorithm can reduce LLM memory usage by 6xarstechnica.com18 points·gmays··3 commentsOpen articleSaveView on HN