There is a patience parameter both in EarlyStopping and ReduceLROnPlateau in Keras. However, the patience parameter for EarlyStopping is 0 meanwhile 10 for ReduceLROnPlateau. Assuming they both monitor val_loss by default, how ReduceLROnPlateau would work if it never has a chance to update itself because of the high patience number?
Then, by logic, patience for EarlyStopping should always be greater than ReduceLROnPlateau, right?
