FEDERATED LEARNING FOR MODEL ADAPTATION

Summary from the TLDR AI newsletter

Researchers have developed a new method called Federated Proxy Fine-Tuning (FedPFT) that improves the adaptation of foundation models for specific tasks while preserving data privacy.