Editing Common Sense in Transformers
Anshita Gupta, Debanjan Mondal, Akshay Krishna Sheshadri, Wenlong Zhao, Xiang Li, Sarah Wiegreffe, Niket Tandon
Abstract
Editing model parameters directly in Transformers makes updating open-source transformer-based models possible without re-training (Meng et al., 2023) . However, these editing methods have only been evaluated on statements about encyclopedic knowledge with a single correct answer. Commonsense knowledge with multiple correct answers, e.g., an apple can be green or red but not transparent, has not been studied but is as essential for enhancing transformers' reliability and usefulness. In this paper, we investigate whether commonsense judgments are causally associated with localized, editable parameters in Transformers, and we provide an affirmative answer. We find that directly applying the MEMIT editing algorithm results in sub-par performance, and propose to improve it for the commonsense domain by varying edit tokens and improving the layer selection strategy, i.e., MEMIT CSK . GPT-2 Large and XL models edited using MEMIT CSK outperform best-fine-tuned baselines by 10.97% and 10.73% F1 scores on PEP3k and 20Q datasets. In addition, we propose a novel evaluation dataset, PROBE SET, that contains unaffected and affected neighborhoods, affected paraphrases, and affected reasoning challenges. MEMIT CSK performs well across the metrics while fine-tuning baselines show significant trade-offs between unaffected and affected metrics. These results suggest a compelling future direction for incorporating feedback about common sense into Transformers through direct model editing. 1
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
20 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Locating and Editing Factual Associations in GPT cites Investigating Gender Bias in Language Models Using Causal Mediation Analysis
- Learning Where to Edit Vision Transformers cites Locating and Editing Factual Associations in GPT
- Boosting Resilience of Large Language Models through Causality-Driven Robust Optimization cites Locating and Editing Factual Associations in GPT
- Boosting Resilience of Large Language Models through Causality-Driven Robust Optimization cites Investigating Gender Bias in Language Models Using Causal Mediation Analysis
- Retrieval Augmentation Reduces Hallucination in Conversation cites Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
- Bidirectional LMs are Better Knowledge Memorizers? A Benchmark for Real-world Knowledge Injection cites Locating and Editing Factual Associations in GPT
- Reasons and Solutions for the Decline in Model Performance after Editing cites Locating and Editing Factual Associations in GPT
- Should We Really Edit Language Models? On the Evaluation of Edited Language Models cites Locating and Editing Factual Associations in GPT
Cited by top-tier papers20
- Should We Really Edit Language Models? On the Evaluation of Edited Language ModelsQi Li, Xiang Liu, Zhenheng Tang, Peijie Dong et al.NeurIPS 2024 · 25 citations
- Reasons and Solutions for the Decline in Model Performance after EditingXiusheng Huang, Jiaxiang Liu, Yequan Wang, Kang LiuNeurIPS 2024 · 13 citations
- Bidirectional LMs are Better Knowledge Memorizers? A Benchmark for Real-world Knowledge InjectionYuwei Zhang, Wenhao Yu, Shangbin Feng, Yifan Zhu et al.ACL 2026 · 7 citations
- Learning Where to Edit Vision TransformersYunqiao Yang, Long-Kai Huang, Shengzhuang Chen, Kede Ma et al.NeurIPS 2024 · 6 citations
- Boosting Resilience of Large Language Models through Causality-Driven Robust OptimizationXiaoling Zhou, Mingjie Zhang, Zhemg Lee, Yuncheng Hua et al.NeurIPS 2025 · 5 citations
Builds on18
- Retrieval-Augmented Generation for Knowledge-Intensive NLP TasksPatrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni et al.NeurIPS 2020 · 19,162 citations
- Locating and Editing Factual Associations in GPTKevin Meng, David Bau, Alex Andonian, Yonatan BelinkovNeurIPS 2022 · 3,415 citations
- Climbing towards NLU: On Meaning, Form, and Understanding in the Age of DataEmily M. Bender, Alexander KollerACL 2020 · 914 citations
- Investigating Gender Bias in Language Models Using Causal Mediation AnalysisJesse Vig, Sebastian Gehrmann, Yonatan Belinkov, Sharon Qian et al.NeurIPS 2020 · 851 citations
- Retrieval Augmentation Reduces Hallucination in ConversationKurt Shuster, Spencer Poff, Moya Chen, Douwe Kiela et al.EMNLP Findings 2021 · 579 citations
Related papers
- Commonsense Knowledge Editing Based on Free-Text in LLMsXiusheng Huang, Yequan Wang, Jun Zhao, Kang LiuEMNLP 2024
- Mass-Editing Memory with Attention in Transformers: A cross-lingual exploration of knowledgeDaniel Mela, Aitor Gonzalez-Agirre, Javier Hernando, Marta VillegasACL Findings 2024 · 3 citations
- One Mask to Rule Them All: On Hidden Facts after Editing and How to Find ThemAli Kholmovaia, Paul Youssef, Nandi Schoots, Christin SeifertACL Findings 2026
- PMET: Precise Model Editing in a TransformerXiaopeng Li, Shasha Li, Shezheng Song, Jing Yang et al.AAAI 2024 · 208 citations
- Mass-Editing Memory in a TransformerKevin Meng, Arnab Sen Sharma, Alex J. Andonian, Yonatan Belinkov et al.ICLR 2023 · 52 citations
