Dynamic Adversarial Fine-Tuning Reorganizes Refusal Geometry 文章

ArXiv CS.CL2026-05-27NEWSen作者: Wenhao Lan, Shan Li, Xinhua Lai, Meiqi Wu, Junbin Yang, Haihua Shen, Yijun Yang

Dynamic Adversarial Fine-Tuning Reorganizes Refusal Geometry · 相关技术