#note how it's the same as the input. That implies
#that the transformations went well
tensor([[[ 0,  1,  2,  3],
         [ 4,  5,  6,  7]],

        [[ 0, -1, -2, -3],
         [-4, -5, -6, -7]]])