This repository presents the code for the paper "Probing Bilingual Guidance for Cross-Lingual Summarization" (NLPCC 2023)

• fairseq version: 0.12.0
• Python version: 3.10.12
• PyTorch version: 2.0.1

To reproduce the results, please follow these steps:

1. Create a conda environment and install the required packages using pip:

   conda env create -n your_env_name python=3.10 -y
   pip install -r requirements.txt

2. Install fairseq in editable mode by navigating to the fairseq directory and running the following command:

   cd fairseq
   pip install --editable ./

3. Download the content of mix_data and mix_data-bin under data/BiNews.

4. Navigate to the corresponding folder and start training via shell script. Take performing English-to-Chinese summarization with Chinese guidance in encoder side as an example:

   cd enzh2zh
   bash enzh2zh_transformer.sh

5. Do inference and evaluation using inference.sh. Note that files2rouge need to be installed for for calculating ROUGE scores.

You may cite our paper as follows:

@inproceedings{zhu2023probing,
  title={Probing Bilingual Guidance for Cross-Lingual Summarization},
  author={Zhu, Dawei and Wu, Wenhao and Li, Sujian},
  booktitle={CCF International Conference on Natural Language Processing and Chinese Computing},
  pages={749--760},
  year={2023},
  organization={Springer}
}

If you have any questions, feel free to open an issue, or contact [email protected];