This course was developed by members of AWS Technical Field Communities (TFC), an AWS community of technical experts. The content is intended to complement our standard training curriculum and augment your AWS learning journey. We are aware some courses have accessibility limitations and are working to address. If you require accommodation, please contact AWS Training and Certification Customer Support.
---
This course introduces Checkpointless training on Amazon SageMaker HyperPod, a revolutionary approach that eliminates traditional checkpoint-based recovery by enabling peer-to-peer state recovery. You'll learn how this technology achieves 80-93% reduction in recovery times (from 15-30+ minutes down to under 2 minutes) and enables 95%+ ML Goodput even at massive scale (1,000-15,000 GPUs).
By the end of this course, you will be able to:
- Understand how checkpointless training addresses the limitations of traditional checkpoint-based recovery approaches and improves ML Goodput
- Identify the core components of checkpointless training that enable sub-2-minute recovery at scale
- Understand the checkpointless training workflow
