Li's picture

Li

WenjieLi

AI & ML interests

None yet

Recent Activity

authored a paper 1 day ago

Why Safeguarded Ships Run Aground? Aligned Large Language Models' Safety Mechanisms Tend to Be Anchored in The Template Region

authored a paper 29 days ago

Sigma: Differential Rescaling of Query, Key and Value for Efficient Language Models

authored a paper 3 months ago

Direct Preference Optimization Using Sparse Feature-Level Constraints

View all activity

Organizations

None yet

Papers 4

arxiv:2502.13946

arxiv:2501.13629

arxiv:2411.07618

arxiv:2408.06941

models

None public yet

datasets

None public yet