how “data” and “target” are choosen in a federated learning? (PySyft)

Question

i can\'t understand how in function train() below, the variable (data, target) are choosen.

def train(args, model, device, federated_train_loader, optimizer, epoc