Static Dropout Layer - Keras

Question

I am interested in making a "dropout like" layer that is static throughout the course of training and testing, in keras. Unlike normal dropout, I only want to sever a certain amount of random weights, not the entire node. I.e. certain neurons cannot connect to others and will never be able to. Is creating a custom layer the easiest way to achieve this? Here is my current progress.

I would also like to be able to visualize the dropout mask.

Answer 1

I don't think you need to edit any source code. Just use seed parameter as given here.