LIBRA: Enabling Workload-aware Multi-dimensional Network Topology Optimization for Distributed Training of Large AI Models

arXiv:2109.11762v2 Announce Type: replace-cross
Abstract: As model sizes in machine learning continue to scale, distributed training is necessary to accommodate model weights within each device and to reduce training time. However, this comes with the expense of increased communication overhead due to the exchange of gradients and activations, which become the critical bottleneck of the end-to-end training process. In this work, we motivate the design of multi-dimensional networks within machine learning systems as a cost-efficient mechanism to enhance overall network …

abstract ai models arxiv communication cs.dc cs.lg distributed enabling however machine machine learning network optimization reduce scale the exchange topology training type