Bilingual Alignment Pre-Training for Zero-Shot Cross-Lingual Transfer

Ziqing Yang, Wentao Ma, Yiming Cui, Jiani Ye, Wanxiang Che, Shijin Wang


Abstract
Multilingual pre-trained models have achieved remarkable performance on cross-lingual transfer learning. Some multilingual models such as mBERT, have been pre-trained on unlabeled corpora, therefore the embeddings of different languages in the models may not be aligned very well. In this paper, we aim to improve the zero-shot cross-lingual transfer performance by proposing a pre-training task named Word-Exchange Aligning Model (WEAM), which uses the statistical alignment information as the prior knowledge to guide cross-lingual word prediction. We evaluate our model on multilingual machine reading comprehension task MLQA and natural language interface task XNLI. The results show that WEAM can significantly improve the zero-shot performance.
Anthology ID:
2021.mrqa-1.10
Volume:
Proceedings of the 3rd Workshop on Machine Reading for Question Answering
Month:
November
Year:
2021
Address:
Punta Cana, Dominican Republic
Editors:
Adam Fisch, Alon Talmor, Danqi Chen, Eunsol Choi, Minjoon Seo, Patrick Lewis, Robin Jia, Sewon Min
Venue:
MRQA
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
100–105
Language:
URL:
https://aclanthology.org/2021.mrqa-1.10
DOI:
10.18653/v1/2021.mrqa-1.10
Bibkey:
Cite (ACL):
Ziqing Yang, Wentao Ma, Yiming Cui, Jiani Ye, Wanxiang Che, and Shijin Wang. 2021. Bilingual Alignment Pre-Training for Zero-Shot Cross-Lingual Transfer. In Proceedings of the 3rd Workshop on Machine Reading for Question Answering, pages 100–105, Punta Cana, Dominican Republic. Association for Computational Linguistics.
Cite (Informal):
Bilingual Alignment Pre-Training for Zero-Shot Cross-Lingual Transfer (Yang et al., MRQA 2021)
Copy Citation:
PDF:
https://preview.aclanthology.org/nschneid-patch-1/2021.mrqa-1.10.pdf