How to perform several different transforms to the same batch image in an efficient way

Hi, I am trying to get several different transformations of the same batch images? I checked this git repository https://github.com/pytorch/vision/issues/157, but I am not sure which one is the best way to do that?