<ul><li>Heterogeneous Federated Learning (HtFL) methods have been developed to enable collaboration across diverse heterogeneous models while tackling data heterogeneity.</li><li>A comprehensive benchmark named Heterogeneous Federated Learning Library (HtFLlib) has been introduced to evaluate and analyze HtFL methods.</li><li>HtFLlib integrates 12 datasets, 40 model architectures, and implementations of 10 representative HtFL methods for research and practical applications.</li><li>The library aims to advance HtFL research and enable broader applications by providing systematic evaluations in terms of accuracy, convergence, computation costs, and communication costs.</li></ul>

HtFLlib: A Comprehensive Heterogeneous Federated Learning Library and Benchmark

Discover more