Overfitting is used to refer to the situation where a model fits well (often too well) to the training set, but fails to generalize that performance to the validation and test sets.