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:
-
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
-
Install
fairseq
in editable mode by navigating to thefairseq
directory and running the following command:cd fairseq pip install --editable ./
-
Download the content of
mix_data
andmix_data-bin
underdata/BiNews
. -
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
-
Do inference and evaluation using
inference.sh
. Note thatfiles2rouge
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 dwzhu@pku.edu.cn;