AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

DRAGON: Guard LLM Unlearning in Context via Negative Detection and Reasoning

Yaxuan Wang, Chris Yuhao Liu, Quan Liu, Jinlong Pang, Wei Wei, Yujia Bao, Yang Liu

ICLR 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning generalization and delta compression · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add CoT guard instruction (vs none) (ablation)DRAGON · TOFU-1% (Llama2-7B-Chat)Deviation Score: 43.9 -> 21.4Improved
use guard-model CoT instead of template CoT (ablation)DRAGON · TOFU-1% (Llama2-7B-Chat)Deviation Score: 33.5 -> 21.4Improved
use guard-model CoT instead of short template CoT (ablation)DRAGON · TOFU-10% (Llama2-7B-Chat)Deviation Score: 40.3 -> 26.5Improved
dual-layer detector vs RoBERTa classifier (ablation)DRAGON · WMDP-biodetection accuracy: 84.2 -> 98.9Improved
dual-layer detector vs GPT-4o classifier (ablation)DRAGON · WMDP-chemdetection accuracy: 100.0 -> 98.3Hurt
DRAGON vs RMU on Yi-34B WMDP-BioDRAGON · WMDP-Bio (Yi-34B-Chat)ProbAcc: 30.6 -> 31.5Hurt
DRAGON vs best baseline on WMDP-Bio (Llama3.1-8B)DRAGON · WMDP-Bio (Llama3.1-8B-Instruct)ProbAcc: 45.1 -> 26.2Improved
DRAGON vs best baseline at TOFU-10%DRAGON · TOFU-10% (Llama2-7B-Chat)Deviation Score: 23.7 -> 26.5Hurt
DRAGON vs gradient-based baselines at TOFU-5%DRAGON · TOFU-5% (Llama2-7B-Chat)Model utility: 0.0 -> 0.6337Improved
DRAGON vs NPO-RT in continual unlearningDRAGON · TOFU continual (Llama2-7B-Chat)DDS: 0.6621 -> 0.2494Improved
scale base model size and use instruct variants (ablation)DRAGON · WMDP/TOFU (Qwen2.5)ROUGE-L gapImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy