Geng Hong

You Can't Eat Your Cake and Have It Too: The Performance Degradation of LLMs with Jailbreak Defense

2025 · arXiv (Cornell University)

With the rise of generative large language models (LLMs) like LLaMA and ChatGPT, these models have significantly transformed daily life and work by providing advanced insights. However, as jailbreak attacks continue to circumvent built-in safety …

SmartSight: Mitigating Hallucination in Video-LLMs Without Compromising Video Understanding via Temporal Attention Collapse

2026 · Proceedings of the AAAI Conference on Artificial Intelligence

Despite Video Large Language Models (Video-LLMs) having rapidly advanced in recent years, perceptual hallucinations pose a substantial safety risk, which severely restricts their real-world applicability. While several methods for hallucination mitigation have been proposed, they …