COSY

This repository contains the code for our ACL 2021 paper:

• Sicheng Yu, Hao Zhang, Yulei Niu, Qianru Sun, Jing Jiang "COSY: COunterfactual SYntax for Cross-Lingual Understanding" (https://aclanthology.org/2021.acl-long.48/)

Requirement

• torch 1.2.0
• stanza 1.0.1
• transformers 3.0.2
• apex 0.1
• tensorboardX 1.8
• prettytable 0.7.2

Cross-lingual Question Answering (MLQA and XQuAD)

Preprocessing

• Step 1: Download training dataset SQuAD via this link (https://drive.google.com/file/d/10STENBjVF3XaCKvzBudtC32B4RWzOwyg/view?usp=sharing) (we rename the file name), and store them in directory /XQA/data/train.
• Step 2: Extract the syntax and process the example

cd XQA/src
bash preprocess.sh

Some tips in extracting the syntax: (1) Do not turn on the --fp16 flag since fp16 is not supported by stanza. (2) Sometimes stanza will collapse and the program keeps printing for all samples, you need to stop the processing, remove the corrupted file and restart the program. (3) If your extracting syntax is not working, feel free to drop me an email and I can share the syntax files.

Training and Inference

cd XQA/src
bash run.sh

QA model trained by me

You can download the model trained by us (with mBERT). You can find the results we reported in our paper. (https://drive.google.com/drive/folders/1SR1mnssCugo81LuucykFixl00LmO347T?usp=sharing)

Recently I am quite busy with the internship and deadline thus the code is still a bit messy. I will come back for the refinement later.

Citation

If you feel this project helpful to your research, please cite our work.

@inproceedings{yu-etal-2021-cosy,
    title = "{COSY}: {CO}unterfactual {SY}ntax for Cross-Lingual Understanding",
    author = "Yu, Sicheng  and
      Zhang, Hao  and
      Niu, Yulei  and
      Sun, Qianru  and
      Jiang, Jing",
    booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)",
    month = aug,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.acl-long.48",
    doi = "10.18653/v1/2021.acl-long.48",
    pages = "577--589",
}