In the domain of sequence modelling, Recurrent Neural Networks (RNN) have
been capable of achieving impressive results in a variety of application areas
including visual question answering, part-of-speech tagging and machine
translation. However this success in modelling short term dependenci