This is neat! I often find that the structure introduced by doing this makes it easier to spot some mistakes. One I've used in ML: train & valid for the trainin

This is neat! I often find that the structure introduced by doing this makes it easier to spot some mistakes. One I've used in ML: train & valid for the training and validation sets.