English

Sequence Modeling with Unconstrained Generation Order

Computation and Language 2019-11-04 v1

Abstract

The dominant approach to sequence generation is to produce a sequence in some predefined order, e.g. left to right. In contrast, we propose a more general model that can generate the output sequence by inserting tokens in any arbitrary order. Our model learns decoding order as a result of its training procedure. Our experiments show that this model is superior to fixed order models on a number of sequence generation tasks, such as Machine Translation, Image-to-LaTeX and Image Captioning.

Keywords

Cite

@article{arxiv.1911.00176,
  title  = {Sequence Modeling with Unconstrained Generation Order},
  author = {Dmitrii Emelianenko and Elena Voita and Pavel Serdyukov},
  journal= {arXiv preprint arXiv:1911.00176},
  year   = {2019}
}

Comments

Camera-ready version for NeurIPS2019

R2 v1 2026-06-23T12:01:47.638Z