EarlyStopping is not early stopping #3
Description
I am new to Tensorflow so I could be wrong, but is it correct to say that the EarlyStopping implemented in this repo is confusing?
The only stop condition I can find is if step > max_steps:, whereas "early stopping" seems to suggest the training is stopped when the training or validation loss is no longer declining.