MMODELYST
People/Zhen Tan
PPL

Zhen Tan

Papers (3)
Measuring Real-World Prompt Injection Attacks in LLM-based Resume ScreeningDetecting Unfaithful Chain-of-Thought via Circuit-Guided Internal-External DiscrepancyTo See is Not to Learn: Protecting Multimodal Data from Unauthorized Fine-Tuning of Large Vision-Language Model