I would like to know the easiest way to create a model, broadcast it with tensorflow federated, run a cycle and collect the weights returned by clients without aggregating t
