作者位次 #9/16OSDI 2026Oral· Operating Systems & Distributed Systems
TrainMover: An Interruption-Resilient Runtime for ML Training
ChonLam Lao · Jiaqi Gao · Jiamin Cao · Zhipeng Zhang · Pengcheng Zhang · Jiangfei Duan · Zhilong Zheng · Yu Guan · Yichi Xu · Yong Li · Zhengping Qian · Aditya Akella · Minlan Yu · Ennan Zhai · Dennis Cai · Jingren Zhou
摘要
Large-scale ML training jobs are frequently interrupted by hardware and software anomalies, failures, and management events. Existing solutions like checkpoint-restart or runtime reconfiguration suffer from long downtimes and degraded performance. We present TrainMover, a resilient LLM training r...