Baughn comments on [Link]: The Unreasonable Effectiveness of Recurrent Neural Networks