Zephon ensures reliable data loading despite system changes
Zephon: Elastic Determinism for Online, Stateful Foundation Model Data Loading Pipelines
Machine LearningArtificial IntelligenceDatabases
Summary
Training large AI models requires loading data in a way that is perfectly repeatable so researchers can trust their results. However, data processing pipelines often change due to hardware differences or interruptions, making exact repetition hard. The authors created Zephon, a system that keeps data loading consistent even when the number of GPUs changes or training stops and restarts. Zephon works by organizing data streams so that ordering decisions are saved and parallel processing can happen safely. This lets researchers run complex, stateful data workflows online without losing reproducibility.
What this means in practice
- •For foundation model engineers: Maintain exact training dataset order across runs even with changing GPU resources and interruptions.
- •For vision-language model developers: Enable efficient and reproducible training on complex video-text datasets without offline data preparation.
Authors
Maximilian Böther, Josh Wills, Ties Robroek, Sonnet Xu, Paul Burstein, Daniel Zayas, Cody Blakeney, Siddharth Joshi, Haoli Yin, Rishabh Adiga, Haakon Mongstad, Luke Merrick, Pratyush Maini, Ari Morcos, Matthew Leavitt, Ana Klimovic, Bogdan Gaza
Abstract
Deterministic data loading is important for foundation model development: model researchers need confidence that differences they observe across costly ablations are caused by the parameter they changed rather than non-determinism in the training data sequence. The data loader must provide elastic determinism, i.e., a deterministic sequence of global training data batches despite changes to the GPU topology across runs (e.g., due to GPU scarcity), frequent checkpoint-resume cycles, and different data processing execution backends. Achieving this is difficult because modern foundation model data pipelines tokenize, pack, and mix samples online, introducing stateful n-to-m transformations that break sample indexing. Existing data loaders largely assume indexable 1-to-1 pipelines, and the common workaround of offline materialization is expensive and, for some modalities such as video, infeasible. We present Zephon, a data loader for foundation models that supports online, stateful pipelines while providing elastic determinism and efficient resumption from checkpoints. It partitions the global stream into topology-independent lanes, serializes ordering decisions while parallelizing stateless work on interchangeable backends, and checkpoints only bounded in-flight state so recovery cost does not grow with training progress. We evaluate Zephon on text and vision-language workloads and show that it achieves competitive throughput while providing a combination of guarantees that no existing loader offers for online, stateful pipelines.