Google's TurboQuant AI-compression algorithm can reduce LLM memory usage by 6x

by geoffbp | View on Hacker News