In my research and learning of Federated Learning, I noticed that most of the researchers are discussing the convergence rates of the models.
Why is that so important and how it can benefit the model discussion?
and if the model fast converged in the first rounds, would that be a bad indication for the model?
