What AI says about FedLess

FedLess is an extended framework for secure and scalable federated learning that runs on serverless computing, supporting CPU and GPU clients and multiple training strategies such as FedAvg, FedProx, Scaffold, FedBuff, FedLessScan, and Apodotiko. Its architecture includes an authentication server, a central FedLess database for client info and logs, a FedLess controller to coordinate training, serverless client functions that perform local training, aggregator functions that combine client updates, and mock components for local development. The project enables distributed model training across clients by coordinating serverless components to orchestrate rounds of federated learning.