Saved in:
Bibliographic Details
Main Authors: Fu, Jinhu, Bai, Yan, He, Longzhu, Lou, Yihang, Zhao, Yanxiao, Sun, Li, Su, Sen
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.05540
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914451409600512
author Fu, Jinhu
Bai, Yan
He, Longzhu
Lou, Yihang
Zhao, Yanxiao
Sun, Li
Su, Sen
author_facet Fu, Jinhu
Bai, Yan
He, Longzhu
Lou, Yihang
Zhao, Yanxiao
Sun, Li
Su, Sen
contents Large language models (LLMs) can effectively handle outdated information through knowledge editing. However, current approaches face two key limitations: (I) Poor generalization: Most approaches rigidly inject new knowledge without ensuring that the model can use it effectively to solve practical problems. (II) Narrow scope: Current methods focus primarily on structured fact triples, overlooking the diverse unstructured forms of factual information (e.g., news, articles) prevalent in real-world contexts. To address these challenges, we propose a new paradigm: teaching LLMs to edit knowledge via Chain of Thoughts (CoTs) reasoning (CoT2Edit). We first leverage language model agents for both structured and unstructured edited data to generate CoTs, building high-quality instruction data. The model is then trained to reason over edited knowledge through supervised fine-tuning (SFT) and Group Relative Policy Optimization (GRPO). At inference time, we integrate Retrieval-Augmented Generation (RAG) to dynamically retrieve relevant edited facts for real-time knowledge editing. Experimental results demonstrate that our method achieves strong generalization across six diverse knowledge editing scenarios with just a single round of training on three open-source language models. The codes are available at https://github.com/FredJDean/CoT2Edit.
format Preprint
id arxiv_https___arxiv_org_abs_2604_05540
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Learning to Edit Knowledge via Instruction-based Chain-of-Thought Prompting
Fu, Jinhu
Bai, Yan
He, Longzhu
Lou, Yihang
Zhao, Yanxiao
Sun, Li
Su, Sen
Computation and Language
Large language models (LLMs) can effectively handle outdated information through knowledge editing. However, current approaches face two key limitations: (I) Poor generalization: Most approaches rigidly inject new knowledge without ensuring that the model can use it effectively to solve practical problems. (II) Narrow scope: Current methods focus primarily on structured fact triples, overlooking the diverse unstructured forms of factual information (e.g., news, articles) prevalent in real-world contexts. To address these challenges, we propose a new paradigm: teaching LLMs to edit knowledge via Chain of Thoughts (CoTs) reasoning (CoT2Edit). We first leverage language model agents for both structured and unstructured edited data to generate CoTs, building high-quality instruction data. The model is then trained to reason over edited knowledge through supervised fine-tuning (SFT) and Group Relative Policy Optimization (GRPO). At inference time, we integrate Retrieval-Augmented Generation (RAG) to dynamically retrieve relevant edited facts for real-time knowledge editing. Experimental results demonstrate that our method achieves strong generalization across six diverse knowledge editing scenarios with just a single round of training on three open-source language models. The codes are available at https://github.com/FredJDean/CoT2Edit.
title Learning to Edit Knowledge via Instruction-based Chain-of-Thought Prompting
topic Computation and Language
url https://arxiv.org/abs/2604.05540