Communication bottlenecks severely hinder the scalability of distributed neural network training, particularly in high-performance computing (HPC) environments. We introduce AB-training, a novel data-parallel method that leverages low-rank representations and independent training groups to significantly reduce communication overhead. Our experiments demonstrate an average reduction in network traffic of approximately 70.31\% across various scaling scenarios, increasing the training potential of communication-constrained systems and accelerating convergence at scale. AB-training also exhibits a pronounced regularization effect at smaller scales, leading to improved generalization while maintaining or even reducing training time. We achieve a remarkable 44.14 : 1 compression ratio on VGG16 trained on CIFAR-10 with minimal accuracy loss, and outperform traditional data parallel training by 1.55\% on ResNet-50 trained on ImageNet-2012. While AB-training is promising, our findings also reveal that large batch effects persist even in low-rank regimes, underscoring the need for further research into optimized update mechanisms for massively distributed training.
@article{arxiv.2405.01067,
title = {AB-Training: A Communication-Efficient Approach for Distributed Low-Rank Learning},
author = {Daniel Coquelin and Katherina Flügel and Marie Weiel and Nicholas Kiefer and Muhammed Öz and Charlotte Debus and Achim Streit and Markus Götz},
journal= {arXiv preprint arXiv:2405.01067},
year = {2024}
}