Does Pre-training Induce Systematic Inference? How Masked Language Models Acquire Commonsense Knowledge

Ian Porada, Alessandro Sordoni, Jackie Cheung


Abstract
Transformer models pre-trained with a masked-language-modeling objective (e.g., BERT) encode commonsense knowledge as evidenced by behavioral probes; however, the extent to which this knowledge is acquired by systematic inference over the semantics of the pre-training corpora is an open question. To answer this question, we selectively inject verbalized knowledge into the pre-training minibatches of BERT and evaluate how well the model generalizes to supported inferences after pre-training on the injected knowledge. We find generalization does not improve over the course of pre-training BERT from scratch, suggesting that commonsense knowledge is acquired from surface-level, co-occurrence patterns rather than induced, systematic reasoning.
Anthology ID:
2022.naacl-main.337
Volume:
Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies
Month:
July
Year:
2022
Address:
Seattle, United States
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
4550–4557
Language:
URL:
https://aclanthology.org/2022.naacl-main.337
DOI:
10.18653/v1/2022.naacl-main.337
Bibkey:
Cite (ACL):
Ian Porada, Alessandro Sordoni, and Jackie Cheung. 2022. Does Pre-training Induce Systematic Inference? How Masked Language Models Acquire Commonsense Knowledge. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4550–4557, Seattle, United States. Association for Computational Linguistics.
Cite (Informal):
Does Pre-training Induce Systematic Inference? How Masked Language Models Acquire Commonsense Knowledge (Porada et al., NAACL 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.naacl-main.337.pdf
Software:
 2022.naacl-main.337.software.zip