{"ID":2885496,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.04012","arxiv_id":"2508.04012","title":"EMSEdit: Efficient Multi-Step Meta-Learning-based Model Editing","abstract":"Large Language Models (LLMs) power numerous AI applications, yet updating their knowledge remains costly. Model editing provides a lightweight alternative through targeted parameter modifications, with meta-learning-based model editing (MLME) demonstrating strong effectiveness and efficiency. However, we find that MLME struggles in low-data regimes and incurs high training costs due to the use of KL divergence. To address these issues, we propose $\\textbf{E}$fficient $\\textbf{M}$ulti-$\\textbf{S}$tep $\\textbf{Edit (EMSEdit)}$, which leverages multi-step backpropagation (MSBP) to effectively capture gradient-activation mapping patterns within editing samples, performs multi-step edits per sample to enhance editing performance under limited data, and introduces norm-based regularization to preserve unedited knowledge while improving training efficiency. Experiments on two datasets and three LLMs show that EMSEdit consistently outperforms state-of-the-art methods in both sequential and batch editing. Moreover, MSBP can be seamlessly integrated into existing approaches to yield additional performance gains. Further experiments on a multi-hop reasoning editing task demonstrate EMSEdit's robustness in handling complex edits, while ablation studies validate the contribution of each design component. Our code is available at https://github.com/xpq-tech/emsedit.","short_abstract":"Large Language Models (LLMs) power numerous AI applications, yet updating their knowledge remains costly. Model editing provides a lightweight alternative through targeted parameter modifications, with meta-learning-based model editing (MLME) demonstrating strong effectiveness and efficiency. However, we find that MLME...","url_abs":"https://arxiv.org/abs/2508.04012","url_pdf":"https://arxiv.org/pdf/2508.04012v3","authors":"[\"Xiaopeng Li\",\"Shasha Li\",\"Xi Wang\",\"Shezheng Song\",\"Bin Ji\",\"Shangwen Wang\",\"Jun Ma\",\"Xiaodong Liu\",\"Mina Liu\",\"Jie Yu\"]","published":"2025-08-06T01:54:58Z","proceeding":"cs.CL","tasks":"[\"cs.CL\",\"cs.AI\",\"cs.LG\"]","methods":"[\"Large Language Model\",\"Language Model\"]","has_code":false,"code_links":[{"ID":611199,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2885496,"paper_url":"https://arxiv.org/abs/2508.04012","paper_title":"EMSEdit: Efficient Multi-Step Meta-Learning-based Model Editing","repo_url":"https://github.com/xpq-tech/emsedit","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}