Switch-GLAT: Multilingual Parallel Machine Translation via Code-Switch Decoder
Abstract
Multilingual machine translation aims to develop a single model for multiple language directions. However, existing multilingual models based on Transformer are limited in terms of both translation performance and inference speed. In this paper, we propose switch-GLAT, a non-autoregressive multilingual machine translation model with a code-switch decoder. It can generate contextual code-switched translations for a given source sentence, and perform code-switch back-translation, greatly boosting multilingual translation performance. In addition, its inference is highly efficient thanks to its parallel decoder. Experiments show that our proposed switch-GLAT outperform the multilingual Transformer with as much as 0.74 BLEU improvement and 6.2x faster decoding speed in inference.
Cite
Text
Song et al. "Switch-GLAT: Multilingual Parallel Machine Translation via Code-Switch Decoder." International Conference on Learning Representations, 2022.Markdown
[Song et al. "Switch-GLAT: Multilingual Parallel Machine Translation via Code-Switch Decoder." International Conference on Learning Representations, 2022.](https://mlanthology.org/iclr/2022/song2022iclr-switchglat/)BibTeX
@inproceedings{song2022iclr-switchglat,
title = {{Switch-GLAT: Multilingual Parallel Machine Translation via Code-Switch Decoder}},
author = {Song, Zhenqiao and Zhou, Hao and Qian, Lihua and Xu, Jingjing and Cheng, Shanbo and Wang, Mingxuan and Li, Lei},
booktitle = {International Conference on Learning Representations},
year = {2022},
url = {https://mlanthology.org/iclr/2022/song2022iclr-switchglat/}
}