【论文阅读】{MegaScale}:Scaling Large Language Model Training to More Than 10,000 {GPUs}

【论文阅读】Fluid:Dataset Abstraction and Elastic Acceleration for Cloud-native Deep Learning Training Jobs