Once-for-All

Efficient Split-Mix Federated Learning for On-Demand and In-Situ Customization

Efficient and federated learning for heterogeneous clients with different memory sizes