Convolutional Sequence to Sequence Learning

Jonas Gehring, Michael Auli, David Grangier, Denis Yarats, Yann N. Dauphin

(Submitted on 8 May 2017)

The prevalent approach to sequence to sequence learning maps an input sequence to a variable length output sequence via recurrent neural networks. We introduce an architecture based entirely on convolutional neural networks. Compared to recurrent models, computations over all elements can be fully parallelized during training and optimization is easier since the number of non-linearities is fixed and independent of the input length. Our use of gated linear units eases gradient propagation and we equip each decoder layer with a separate attention module. We outperform the accuracy of the deep LSTM setup of Wu et al. (2016) on both WMT'14 English-German and WMT'14 English-French translation at an order of magnitude faster speed, both on GPU and CPU.

Subjects:Computation and Language (cs.CL)Cite as:arXiv:1705.03122 [cs.CL] (or arXiv:1705.03122v1 [cs.CL] for this version)

Submission history

From: Michael Auli [view email] 
[v1] Mon, 8 May 2017 23:25:30 GMT (1489kb,D)