Exploring multi-dimensional hierarchical network topologies for efficient distributed training of trillion parameter dl models
Deep Neural Networks have gained significant attraction due to their wide applicability in
different domains. DNN sizes and training samples are constantly growing, making training …
different domains. DNN sizes and training samples are constantly growing, making training …
Exploring Multi-dimensional Hierarchical Network Topologies for Efficient Distributed Training of Trillion Parameter DL Models
W Won, S Rashidi, S Srinivasan, T Krishna - arXiv e-prints, 2021 - ui.adsabs.harvard.edu
Abstract Deep Neural Networks have gained significant attraction due to their wide
applicability in different domains. DNN sizes and training samples are constantly growing …
applicability in different domains. DNN sizes and training samples are constantly growing …