评分依据:Novel MoE approach addressing expert routing efficiency at scale. Strong relevance to frontier model architecture design.
Expert Upcycling: Shifting the Compute-Efficient Frontier of Mixture-of-Experts
发布
采集
行业动态 7.5 分
— Novel MoE approach addressing expert routing efficiency at scale. Strong relevance to frontier model architecture design. 原文: arXiv