Zhiyuan Wang
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
When Safety Fails Collectively: Robust Alignment Against Harmful Fine-Tuning via Joint Weight Perturbations
According to the abstract, Zhiyuan Wang, Yuezhi Zhou, Zhu Zhang and Zhixing Tan observe that separate parts of a harmful weight perturbation have limited effect alone but together raise harmfulness beyond their sum. They propose Joint Weight Perturbations, an alignment-stage defense, and report stronger resistance to harmful fine-tuning.