For you nerds: There is two weights, two biases, and three nodes: input -> hidden -> output it has no activation and takes x and outputs y, the star shows, it does gradient descent
