twoclassmnist
Table of contents
Demo neural network app implemented using nn and nn-utils. Ideally, every neural network should be implemented in separate apps like this and should have at least 1 forward() function.
Functions
forward
This forward() implements the neural network using functions from nn app. To run the neural network, you can use /script in twoclassmnist invoke forward or simply /twoclassmnist forward in any Minecraft command block.
clear_activations
This function clears the intermediate activations created during last forward pass. Although not technically compulsory function, it is nice to have this function in case you want to visualize the neural network block by block after every forward pass.