Probing for Bridging Inference in Transformer Language Models

Onkar Pandit, Yufang Hou


Abstract
We probe pre-trained transformer language models for bridging inference. We first investigate individual attention heads in BERT and observe that attention heads at higher layers prominently focus on bridging relations in-comparison with the lower and middle layers, also, few specific attention heads concentrate consistently on bridging. More importantly, we consider language models as a whole in our second approach where bridging anaphora resolution is formulated as a masked token prediction task (Of-Cloze test). Our formulation produces optimistic results without any fine-tuning, which indicates that pre-trained language models substantially capture bridging inference. Our further investigation shows that the distance between anaphor-antecedent and the context provided to language models play an important role in the inference.
Anthology ID:
2021.naacl-main.327
Volume:
Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies
Month:
June
Year:
2021
Address:
Online
Editors:
Kristina Toutanova, Anna Rumshisky, Luke Zettlemoyer, Dilek Hakkani-Tur, Iz Beltagy, Steven Bethard, Ryan Cotterell, Tanmoy Chakraborty, Yichao Zhou
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
4153–4163
Language:
URL:
https://aclanthology.org/2021.naacl-main.327
DOI:
10.18653/v1/2021.naacl-main.327
Bibkey:
Cite (ACL):
Onkar Pandit and Yufang Hou. 2021. Probing for Bridging Inference in Transformer Language Models. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4153–4163, Online. Association for Computational Linguistics.
Cite (Informal):
Probing for Bridging Inference in Transformer Language Models (Pandit & Hou, NAACL 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.naacl-main.327.pdf
Video:
 https://aclanthology.org/2021.naacl-main.327.mp4
Code
 oapandit/probBertForbridging
Data
ISNotes