there is no chage to the dataset even after random affine (pytorch)

Question

I downloaded two version of mnist dataset: one without any transformation, one with random affinement and normalization.

w/o transformation code:

mnist_train = torchvision.datasets.MNIST(root='MNIST_data/', 
                          train=True, 
                          transform=transforms.ToTensor(), 
                          download=True)

with transformation code:

train_transforms = transforms.Compose(
    [transforms.RandomAffine(degrees=30),
     transforms.ToTensor(),
     transforms.Normalize(mean=(0.5,), std=(0.5,))])

augmented_data = torchvision.datasets.MNIST(root='aug_data/',
                                        train=True,
                                        transform=train_transforms,
                                        download=True)

I printed several samples from those two, but the results were all identical. can anybody figure out the cause??

thank you