Reimplementation of convolutional sequence to sequence learning
Recurrent Neural Network has proven to show good results in terms of natural language processing or in regards with data in sequence. Instead of using recurrent neural network, this paper uses convolutional to learn the mapping between sequence to sequence. The computation time for convolution will be faster than of recurrent because of its power in parallelism. To try out convolutional network for sequence to sequence learning. I decided to try and reimplement the model.