full_dataset = datasets.ImageFolder(os.path.join(root, ‘Imagenes_categoria’), transform=train_transform)
ttv=0.7 #tasa de entrenamiento y test
size_t = int(ttv * len(full_dataset))
size_v = len(full_dataset) - size_t
train_data, test_data= torch.utils.data.random_split(full_dataset, [size_t, size_v])
how can I save an image folder after the split, the train_data folder on my disk, and then apply validations to the epochs?