On the 5-way 1-shot task of Omniglot there are two metrics in the CSV results file: categorical_accuracy and val_1-shot_5-way_acc.
I want to confirm which metric is used to evaluate the performace of the model on the dataset: categorical_accuracy or
val_1-shot_5-way_acc?