Dongcheng Zhao

Safety Instincts: LLMs Learn to Trust Their Internal Compass for Self-Defense

Fine-Grained Safety Neurons with Training-Free Continual Projection to Reduce LLM Fine Tuning Risks

Harnessing task overload for scalable jailbreak attacks on large language models

Learning the Plasticity: Plasticity-Driven Learning Framework in Spiking Neural Networks

Jailbreak Antidote: Runtime Safety-Utility Balance via Sparse Representation Adjustment in Large Language Models

Harnessing Task Overload for Scalable Jailbreak Attacks on Large Language Models