Exploring the Role of BERT Token Representations to Explain Sentence Probing Results

Hosein Mohebbi, Ali Modarressi, Mohammad Taher Pilehvar


Abstract
Several studies have been carried out on revealing linguistic features captured by BERT. This is usually achieved by training a diagnostic classifier on the representations obtained from different layers of BERT. The subsequent classification accuracy is then interpreted as the ability of the model in encoding the corresponding linguistic property. Despite providing insights, these studies have left out the potential role of token representations. In this paper, we provide a more in-depth analysis on the representation space of BERT in search for distinct and meaningful subspaces that can explain the reasons behind these probing results. Based on a set of probing tasks and with the help of attribution methods we show that BERT tends to encode meaningful knowledge in specific token representations (which are often ignored in standard classification setups), allowing the model to detect syntactic and semantic abnormalities, and to distinctively separate grammatical number and tense subspaces.
Anthology ID:
2021.emnlp-main.61
Volume:
Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2021
Address:
Online and Punta Cana, Dominican Republic
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
792–806
Language:
URL:
https://aclanthology.org/2021.emnlp-main.61
DOI:
10.18653/v1/2021.emnlp-main.61
Bibkey:
Cite (ACL):
Hosein Mohebbi, Ali Modarressi, and Mohammad Taher Pilehvar. 2021. Exploring the Role of BERT Token Representations to Explain Sentence Probing Results. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 792–806, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics.
Cite (Informal):
Exploring the Role of BERT Token Representations to Explain Sentence Probing Results (Mohebbi et al., EMNLP 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.emnlp-main.61.pdf
Video:
 https://aclanthology.org/2021.emnlp-main.61.mp4
Code
 hmohebbi/explain-probing-results