loss

FullyConnected.loss(y, y_pred, sample_weight=None)