Handling Network Faults in Distributed AI Training: Failover is Now an Option
Open Access
- 26 April 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 263-278
Abstract
No abstract availableThis publication has 9 references indexed in Scilit:
- ReCycle: Resilient Training of Large DNNs using Pipeline Adaptation2024
- Alibaba HPN: A Data Center Network for Large Language Model Training2024
- BEVFormer v2: Adapting Modern Image Backbones to Bird's-Eye-View Recognition via Perspective Supervision2023
- ACCL: Architecting Highly Scalable Distributed Training Systems With Highly Efficient Collective Communication LibraryIEEE Micro, 2021
- Surviving switch failures in cloud datacentersACM SIGCOMM Computer Communication Review, 2021
- DeepSpeed2020
- PyTorch distributedProceedings of the VLDB Endowment, 2020
- Understanding and Mitigating Packet Corruption in Data Center Networks2017
- Understanding network failures in data centers2011