Explanation of Masked Training

[OliviaOliveiira]

Describe your use-case.

To better understand the concept, use when needed, and avoid when not.

What would you like to see as a solution?

A detailed explanation of how you've implemented it, the way it affects the training and what happens with the background, does it go fully "nan" or it remains but at a much slower "weight"

Have you considered alternatives? List them here.

Only through author's explanation.