Usually when a model overfits, validation loss goes up and training loss goes down from the point of overfitting. But for my case, training loss still goes down but validati
