Prix-LM

Code for paper Prix-LM: Pretraining for Multilingual Knowledge Base Construction.

Requirements

• PyTorch
• Transformers
• wandb
• tqdm

Training and Evaluation

We have released the code for link prediction. Code on other tasks will be released soon.

Link Prediction

Before running link prediction, unzip the evaluation data under the data/link_prediction folder. Run link prediction on our DBpedia test set by:

>> python lp.py --lan language

The lan parameter can take the 9 lanugages in our paper (en, it, fr, de, fi, et, hu, tr, and ja).

This project is supported by by the National Science Foundation of United States Grant IIS 2105329.

@inproceedings{zhou-etal-2022-prix,
    title = "Prix-{LM}: Pretraining for Multilingual Knowledge Base Construction",
    author = "Zhou, Wenxuan  and
      Liu, Fangyu  and
      Vuli{\'c}, Ivan  and
      Collier, Nigel  and
      Chen, Muhao",
    booktitle = "Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    year = "2022",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.acl-long.371",
    pages = "5412--5424"
}