Jul 22 1-bit LLMs could solve AI’s energy demands: « “Imprecise” language models are smaller, speedier—and nearly as accurate. »
