Efficient Language Model Architectures for Differentially Private Federated Learning

arXiv:2403.08100v1 Announce Type: cross
Abstract: Cross-device federated learning (FL) is a technique that trains a model on data distributed across typically millions of edge devices without data leaving the devices. SGD is the standard client optimizer for on device training in cross-device FL, favored for its memory and computational efficiency. However, in centralized training of neural language models, adaptive optimizers are preferred as they offer improved stability and performance. In light of this, we ask if language models can be …

architectures arxiv client computational cs.cr cs.lg data device devices distributed edge edge devices efficiency federated federated learning language memory millions private standard training trains