My solution to the Udacity project of training all Seinfeld Scripts in an RNN to predict a new script based on which character talks first