The model you have trained has a low bias and high variance. How would you deal with it?