Google's TurboQuant AI-compression algorithm can reduce LLM memory usage by 6x(arstechnica.com)
16 pointsby gmays6 hours ago1 comment