Abstract
This paper investigates Who’s Harry Potter (WHP), a pioneering yet insufficiently understood method for LLM unlearning. We explore it in two steps. First, we introduce a new task of LLM targeted unlearning, where given an unlearning target (e.g., a person) and some unlearning documents, we aim to unlearn only the information about the target, rather than everything in the unlearning documents. We further argue that a successful unlearning should satisfy criteria such as not outputting gibberish, not fabricating facts about the unlearning target, and not releasing factual information under jailbreak attacks. Second, we construct a causal intervention framework for targeted unlearning, where the knowledge of the unlearning target is modeled as a confounder between LLM input and output, and the unlearning process as a deconfounding process. This framework justifies and extends WHP, deriving a simple unlearning algorithm that includes WHP as a special case. Experiments on existing and new datasets show that our approach, without explicitly optimizing for the aforementioned criteria, achieves competitive performance in all of them.- Anthology ID:
- 2024.emnlp-main.495
- Volume:
- Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
- Month:
- November
- Year:
- 2024
- Address:
- Miami, Florida, USA
- Editors:
- Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 8708–8731
- Language:
- URL:
- https://preview.aclanthology.org/add_missing_videos/2024.emnlp-main.495/
- DOI:
- 10.18653/v1/2024.emnlp-main.495
- Cite (ACL):
- Yujian Liu, Yang Zhang, Tommi Jaakkola, and Shiyu Chang. 2024. Revisiting Who’s Harry Potter: Towards Targeted Unlearning from a Causal Intervention Perspective. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 8708–8731, Miami, Florida, USA. Association for Computational Linguistics.
- Cite (Informal):
- Revisiting Who’s Harry Potter: Towards Targeted Unlearning from a Causal Intervention Perspective (Liu et al., EMNLP 2024)
- PDF:
- https://preview.aclanthology.org/add_missing_videos/2024.emnlp-main.495.pdf