This paper describes NiuTrans neural machine translation systems of the WMT 2021 news translation tasks. We made submissions to 9 language directions, including English↔{Chinese, Japanese, Russian, Icelandic} and English→Hausa tasks. Our primary systems are built on several effective variants of Transformer, e.g., Transformer-DLCL, ODE-Transformer. We also utilize back-translation, knowledge distillation, post-ensemble, and iterative fine-tuning techniques to enhance the model performance further.
@article{arxiv.2109.10485,
title = {The NiuTrans Machine Translation Systems for WMT21},
author = {Shuhan Zhou and Tao Zhou and Binghao Wei and Yingfeng Luo and Yongyu Mu and Zefan Zhou and Chenglong Wang and Xuanjun Zhou and Chuanhao Lv and Yi Jing and Laohu Wang and Jingnan Zhang and Canan Huang and Zhongxiang Yan and Chi Hu and Bei Li and Tong Xiao and Jingbo Zhu},
journal= {arXiv preprint arXiv:2109.10485},
year = {2021}
}