KLAR-CLC

This repository provides the dataset for:

Lost in Multilinguality: Dissecting Cross-lingual Factual Inconsistency in Transformer Language Models

💡 Introduction

We apply mechanistic interpretability methods to analyze cross-lingual inconsistencies in multilingual masked language models (MLMs). Our findings reveal that MLMs represent knowledge in a shared, language-independent space through most layers, transitioning to language-specific representations only in the final layers. Errors often occur during this transition, leading to incorrect predictions in the target language despite correct answers in others. These insights offer a lightweight and effective strategy for improving factual consistency across languages.

This repository hosts KLAR, the dataset we created for multilingual knowledge probing, covering 17 languages. The accompanying code is available here.

📙 Citation

If you found our work useful for your research, please cite it as follows:

@inproceedings{wang-etal-2025-lost-multilinguality,
    title = "Lost in Multilinguality: Dissecting Cross-lingual Factual Inconsistency in Transformer Language Models",
    author = {Wang, Mingyang and Adel, Heike and Lange, Lukas and Liu, Yihong and Nie, Ercong and Strötgen, Jannik and Schütze, Hinrich},
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.acl-long.253/",
}