Knowledge Neurons in Pretrained Transformers
Damai Dai, Li Dong, Yaru Hao, Zhifang Sui, Baobao Chang, Furu Wei
Abstract
Large-scale pretrained language models are surprisingly good at recalling factual knowledge presented in the training corpus. In this paper, we present preliminary studies on how factual knowledge is stored in pretrained Transformers by introducing the concept of knowledge neurons. Specifically, we examine the fill-in-the-blank cloze task for BERT. Given a relational fact, we propose a knowledge attribution method to identify the neurons that express the fact. We find that the activation of such knowledge neurons is positively correlated to the expression of their corresponding facts. In our case studies, we attempt to leverage knowledge neurons to edit (such as update, and erase) specific factual knowledge without fine-tuning. Our results shed light on understanding the storage of knowledge within pretrained Transformers.- Anthology ID:
- 2022.acl-long.581
- Volume:
- Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
- Month:
- May
- Year:
- 2022
- Address:
- Dublin, Ireland
- Editors:
- Smaranda Muresan, Preslav Nakov, Aline Villavicencio
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 8493–8502
- Language:
- URL:
- https://aclanthology.org/2022.acl-long.581
- DOI:
- 10.18653/v1/2022.acl-long.581
- Cite (ACL):
- Damai Dai, Li Dong, Yaru Hao, Zhifang Sui, Baobao Chang, and Furu Wei. 2022. Knowledge Neurons in Pretrained Transformers. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 8493–8502, Dublin, Ireland. Association for Computational Linguistics.
- Cite (Informal):
- Knowledge Neurons in Pretrained Transformers (Dai et al., ACL 2022)
- PDF:
- https://preview.aclanthology.org/ml4al-ingestion/2022.acl-long.581.pdf
- Code
- hunter-ddm/knowledge-neurons + additional community code