A arstechnica.com Google's TurboQuant AI-compression algorithm can reduce LLM memory usage by 6x 9 points geoffbp 7 months ago 5 comments
