On the Representational Capacity of Recurrent Neural Language Models

Open in new window