I’ve downloaded the EMNIST dataset of letters, and I’ve converted each image to a torch.tensor object with shape torch.size([28, 28]). However, I would like divide the 28*28 image into 7*7 blocks, with each block sized 16.