Model-in-Deep-Learning-with-multiple-outputs

When training a model in Deep Learning with multiple outputs, such as ROC1, ROC2, ROC3, ROC4, ROC5, what does the learning curve show? Is it the average of the learning curves of all 5 outputs?