This is a follow up to my previous post on training this model.