AI Paper Atlas

Research map / Large language models / Alignment and preferences

Revisiting Robustness for LLM Safety Alignment via Selective Geometry Control

Yonghui Yang, Wenjian Tao, Jilong Liu, Xingyu Zhu, Junfeng Fang, Huang Weibiao, Le Wu, Richang Hong, Tat-Seng Chua

ICML 2026 · Large language models › Alignment and preferences · approach: Robust and calibrated DPO · 0 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 token-level ShaPO selective geometry control to DPOLLaMA-3-8B DPO · PKU-SafeRLHF-30K IIDwin rate: 69.19% -> 85.45%Improved
add selective geometry control vs Dr.DPOLLaMA-3-8B · OOD safety averageattack success rate: 2.09% -> 1.37% (ShaPO-T)Improved
switch to reward-level ShaPO-R from token-level ShaPO-TLLaMA-3-8B · OOD safety averageattack success rate: 1.37% -> 0.70%Improved
switch to reward-level ShaPO-R from token-level ShaPO-TQwen2.5-7B · OOD safety averageattack success rate: 0.42% -> 0.55%Hurt
random neuron perturbation instead of no control (ablation)DPO LLaMA-3-8B · OOD average MDattack success rate: 28.78% -> 28.56%No clear effect
uniform geometry control instead of no control (ablation)DPO LLaMA-3-8B · OOD average MDattack success rate: 28.78% -> 26.83%Improved
selective instead of uniform geometry control (ablation)DPO LLaMA-3-8B · OOD average MDattack success rate: 26.83% -> 26.31%Improved
selective instead of no control, IID evaluation (ablation)DPO LLaMA-3-8B · PKU-SafeRLHF-30K IID MDattack success rate: 26.57% -> 27.32%No clear effect
increase selective neuron ratio from 0.01% to 1% (ablation)ShaPO LLaMA-3-8B · OOD average ASRattack success rateImproved
increase selective neuron ratio beyond 1% (ablation)ShaPO LLaMA-3-8B · OOD average ASRattack success rateHurt
add ShaPO-R geometry control on top of reward supervision (DPO) (ablation)DPO+Reward LLaMA-3-8B · OOD average MDattack success rate: 17.33% -> 15.98%Improved
add ShaPO-R geometry control on top of reward supervision (Dr.DPO) (ablation)Dr.DPO+Reward LLaMA-3-8B · OOD average MDattack success rate: 2.25% -> 1.30%Improved

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