Challenge: simple feedforward NN with variable amount of layers.

Dan has done a great series on simple feed forward NN (Toy Neural Network). But it was limited to only 2 layers. it would be nice to see him make a simple feed forward NN but with variable amount of layers. this will be great to see in p5, but even better to see it in python. Sebastian Lague started series where he made these Variable Length NNs but he dropped that series.