Context-Aware Model on Fairseq

The implementation of "Does Multi-Encoder Help? A Case Study on Context-Aware Neural Machine Translation"

This code is based on Fairseq v0.6.2

Requirements and Installation

• PyTorch version >= 1.0.0
• Python version >= 3.6

1. pip3 install -r requirements.txt
2. python3 setup.py develop
3. python3 setup.py install

Prepare Training Data

bash runs/prepare-en2ru.sh

Train

Train transformer baseline

bash runs/train-en2ru.sh baseline

Train context-aware model

bash runs/train-en2ru.sh inside-context

bash runs/train-en2ru.sh outside-context

Train model with gaussian noise

bash runs/train-en2ru.sh gaussian

Infer

bash runs/translate-en2ru.sh baseline

bash runs/translate-en2ru.sh inside-context

bash runs/translate-en2ru.sh outside-context

bash runs/translate-en2ru.sh gaussian

Infer without context

bash runs/translate-en2ru.sh inside-context ignore

bash runs/translate-en2ru.sh outside-context ignore

Citation

@inproceedings{li-etal-2020-multi,
    title = "Does Multi-Encoder Help? A Case Study on Context-Aware Neural Machine Translation",
    author = "Li, Bei  and
      Liu, Hui  and
      Wang, Ziyang  and
      Jiang, Yufan  and
      Xiao, Tong  and
      Zhu, Jingbo  and
      Liu, Tongran  and
      li, changliang",
    booktitle = "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics",
    month = jul,
    year = "2020",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2020.acl-main.322",
    pages = "3512--3518",
}