How to preprocess datasets.

Thanks for your great work. For train domain yoh have given the preprocessing code, but for test domain how to obtain slices and save as npz, can you open source more detailed processing code, thanks.
