Dynamic Retriever for In-Context Knowledge Editing via Policy Optimization

2025-10-24 · Mahmud Wasif Nafee, Maiqi Jiang, Haipeng Chen, Yanfu Zhang arxiv

Large language models (LLMs) excel at factual recall yet still propagate stale or incorrect knowledge. In-context knowledge editing offers a gradient-free remedy suitable for black-box APIs, but current editors rely on static demonstration sets chosen by surface-level similarity, leading to two persistent obstacles: (i) a quantity-quality trade-off, and (ii) lack of adaptivity to task difficulty. We address these issues by dynamically selecting supporting demonstrations according to their utility for the edit. We propose Dynamic Retriever for In-Context Knowledge Editing (DR-IKE), a lightweight framework that (1) trains a BERT retriever with REINFORCE to rank demonstrations by editing reward, and (2) employs a learnable threshold to prune low-value examples, shortening the prompt when the edit is easy and expanding it when the task is hard. DR-IKE performs editing without modifying model weights, relying solely on forward passes for compatibility with black-box LLMs. On the COUNTERFACT benchmark, it improves edit success by up to 17.1%, reduces latency by 41.6%, and preserves accuracy on unrelated queries, demonstrating scalable and adaptive knowledge editing. The code is available at https://github.com/mwnafee/DR-IKE .

📄 PDF Abstract BibTeX arXiv:2510.21059

Code (0)

등록된 구현이 없습니다.

Tasks

knowledge editing

Conflict-Aware Retriever Editing for Knowledge Injection Attacks on LLM-Based RAG Systems

2026-06-16 · Xinru Liu, Xianglong Zhang, Di Cai, Zhumin Chen 외 arxiv

Injecting malicious knowledge into retrieval-augmented generation (RAG) systems can manipulate retrieved evidence and mislead downstream generation, posing a serious security threat for AI applications. Existing RAG inje…

Dynamic Injection of Entity Knowledge into Dense Retrievers

2025-07-05 · Ikuya Yamada, Ryokan Ri, Takeshi Kojima, Yusuke Iwasawa 외 arxiv

Dense retrievers often struggle with queries involving less-frequent entities due to their limited entity knowledge. We propose the Knowledgeable Passage Retriever (KPR), a BERT-based retriever enhanced with a context-en…

Decoupled Reasoning with Implicit Fact Tokens (DRIFT): A Dual-Model Framework for Efficient Long-Context Inference

2026-02-10 · Wenxuan Xie, Yujia Wang, Xin Tan, Chaochao Lu 외 arxiv

The integration of extensive, dynamic knowledge into Large Language Models (LLMs) remains a significant challenge due to the inherent entanglement of factual data and reasoning patterns. Existing solutions, ranging from …

knowledge editing

Lifelong Knowledge Editing for LLMs with Retrieval-Augmented Continuous Prompt Learning

2024-05-06 · Qizhou Chen, Taolin Zhang, Xiaofeng He, Dongyang Li 외

Model editing aims to correct outdated or erroneous knowledge in large language models (LLMs) without the need for costly retraining. Lifelong model editing is the most challenging task that caters to the continuous edit…

knowledge editing Lifelong learning Model Editing Prompt Learning+1

Disabling Self-Correction in Retrieval-Augmented Generation via Stealthy Retriever Poisoning

2025-08-27 · Yanbo Dai, Zhenlan Ji, Zongjie Li, Kuan Li 외 arxiv

Retrieval-Augmented Generation (RAG) has become a standard approach for improving the reliability of large language models (LLMs). Prior work demonstrates the vulnerability of RAG systems by misleading them into generati…