Why only utilize the test dataset for both training and testing？

[yujunzhen]

@caoyunkang Hello,a little question please. Why not separate the training and testing datasets?

`
class BaseDataset(data.Dataset):
def init(self, clsnames, transform, target_transform, root, aug_rate=0., training=True):
self.root = root
self.transform = transform
self.target_transform = target_transform
self.aug_rate = aug_rate
self.training = training
self.data_all = []
self.cls_names = clsnames
solver = DataSolver(root, clsnames)
meta_info = solver.run()

    self.meta_info = meta_info['test']  # Only utilize the test dataset for both training and testing 
    for cls_name in self.cls_names:
        self.data_all.extend(self.meta_info[cls_name])

    self.length = len(self.data_all)

`

[caoyunkang]

We followed the default settings in the APRIL-GAN repository. Specifically, we used the testing data from one dataset as the training data and then tested the model on another dataset.