Convert the 1-D integer tensor labels into a one-hot encoded matrix with num_classes columns: row i is all zeros except a 1 at column labels[i].

Input

labels = tensor([0, 2, 1, 3])
num_classes = 4

Output

tensor([[1, 0, 0, 0],
        [0, 0, 1, 0],
        [0, 1, 0, 0],
        [0, 0, 0, 1]])