DRAGON: Guard LLM Unlearning in Context via Negative Detection and Reasoning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914149720653824 |
|---|---|
| author | Wang, Yaxuan Liu, Chris Yuhao Liu, Quan Pang, Jinglong Wei, Wei Bao, Yujia Liu, Yang |
| author_facet | Wang, Yaxuan Liu, Chris Yuhao Liu, Quan Pang, Jinglong Wei, Wei Bao, Yujia Liu, Yang |
| contents | Unlearning in Large Language Models (LLMs) is crucial for protecting private data and removing harmful knowledge. Most existing approaches rely on fine-tuning to balance unlearning efficiency with general language capabilities. However, these methods typically require training or access to retain data, which is often unavailable in real world scenarios. Although these methods can perform well when both forget and retain data are available, few works have demonstrated equivalent capability in more practical, data-limited scenarios. To overcome these limitations, we propose Detect-Reasoning Augmented GeneratiON (DRAGON), a systematic, reasoning-based framework that utilizes in-context chain-of-thought (CoT) instructions to guard deployed LLMs before inference. Instead of modifying the base model, DRAGON leverages the inherent instruction-following ability of LLMs and introduces a lightweight detection module to identify forget-worthy prompts without any retain data. These are then routed through a dedicated CoT guard model to enforce safe and accurate in-context intervention. To robustly evaluate unlearning performance, we introduce novel metrics for unlearning performance and the continual unlearning setting. Extensive experiments across three representative unlearning tasks validate the effectiveness of DRAGON, demonstrating its strong unlearning capability, scalability, and applicability in practical scenarios. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_05784 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | DRAGON: Guard LLM Unlearning in Context via Negative Detection and Reasoning Wang, Yaxuan Liu, Chris Yuhao Liu, Quan Pang, Jinglong Wei, Wei Bao, Yujia Liu, Yang Computation and Language Artificial Intelligence Machine Learning Unlearning in Large Language Models (LLMs) is crucial for protecting private data and removing harmful knowledge. Most existing approaches rely on fine-tuning to balance unlearning efficiency with general language capabilities. However, these methods typically require training or access to retain data, which is often unavailable in real world scenarios. Although these methods can perform well when both forget and retain data are available, few works have demonstrated equivalent capability in more practical, data-limited scenarios. To overcome these limitations, we propose Detect-Reasoning Augmented GeneratiON (DRAGON), a systematic, reasoning-based framework that utilizes in-context chain-of-thought (CoT) instructions to guard deployed LLMs before inference. Instead of modifying the base model, DRAGON leverages the inherent instruction-following ability of LLMs and introduces a lightweight detection module to identify forget-worthy prompts without any retain data. These are then routed through a dedicated CoT guard model to enforce safe and accurate in-context intervention. To robustly evaluate unlearning performance, we introduce novel metrics for unlearning performance and the continual unlearning setting. Extensive experiments across three representative unlearning tasks validate the effectiveness of DRAGON, demonstrating its strong unlearning capability, scalability, and applicability in practical scenarios. |
| title | DRAGON: Guard LLM Unlearning in Context via Negative Detection and Reasoning |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2511.05784 |