A real MLP it kinda works specs: 1 hidden layer! 3 neurons! tanh activation (clanky) real accuracy! real forward pass and backpropagation! worst formatted code! 1 visual cube thingy! no curves! (I hate the pen extension) training on sine(input) somehow improves over time! for some documentation see inside. If you outsome reason wanna use my code please credit me.
wikipedia for math equations. [https://en.wikipedia.org/wiki/Multilayer_perceptron]
