This is the blueprint for a MLP layer - Using the sprite (MLPLayer) inside, one can create (relatively low-level) MLPs. This uses sprite-only variables to create an easily copy-and-past-able while passing information through the global variables. Keep in mind that a smoothed version of ReLU is used, but can be changed (one must also change it in other feed-forward functions and the gradients in the needed gradient-storing feed-forward function(s)). Feel free to use with credit, even in small projects, for if someone else would like this as a blueprint (or link to a more useful copy).
All code by me.
