Softmax regression (or multinomial logistic regression) is a generalization of logistic regression to the case where we want to handle multiple classes.

This notebook implements the algorithm from scratch with "l2" regularization and early stopping and trains the model on the iris dataset.