In this exercise you need to implement a $3-$layer MLP model $($one input layer, one hidden layer with tanh activation and one output layer$)$ in any platform, which will be used to classify the images from the MNIST dataset.