Validation is done using the test set

[SWaltmann]

In model.py, the train method of EfficientCapsNet includes the following line:

dataset_train, dataset_val = dataset.get_tf_data()

However, get_tf_data() simply calls generate_tf_data(), which augments and returns the test dataset.

Was this done on purpose? My understanding is that the test set should remain unseen until the actual test.

Cheers!

Stijn