ScaleAcross Explorer: Exploring Communication Optimization for Scale-Across AI Model Training 事件

ScaleAcross Explorer: Exploring Communication Optimization for Scale-Across AI Model Training arXiv:2605.24326v1 Announce Type: cross Abstract: The rapid scaling of large language model training requires distributing GPU resources across multiple data center buildings and regions. We refer to such paradigm as "scale-across" training. As infrastructure expands, the system design space becomes increasingly intricate, encompassing new model architectures, hardware heterogeneity, and evolving commu