English

Off-the-Shelf Unsupervised NMT

Computation and Language 2018-11-07 v1

Abstract

We frame unsupervised machine translation (MT) in the context of multi-task learning (MTL), combining insights from both directions. We leverage off-the-shelf neural MT architectures to train unsupervised MT models with no parallel data and show that such models can achieve reasonably good performance, competitive with models purpose-built for unsupervised MT. Finally, we propose improvements that allow us to apply our models to English-Turkish, a truly low-resource language pair.

Keywords

Cite

@article{arxiv.1811.02278,
  title  = {Off-the-Shelf Unsupervised NMT},
  author = {Chris Hokamp and Sebastian Ruder and John Glover},
  journal= {arXiv preprint arXiv:1811.02278},
  year   = {2018}
}
R2 v1 2026-06-23T05:05:59.075Z