Haebin Seong

D2E: Scaling Vision-Action Pretraining on Desktop Data for Transfer to Embodied AI

BiasJailbreak:Analyzing Ethical Biases and Jailbreak Vulnerabilities in Large Language Models

FedSVD: Adaptive Orthogonalization for Private Federated Learning with LoRA

HarmAug: Effective Data Augmentation for Knowledge Distillation of Safety Guard Models

Do LLMs Have Political Correctness? Analyzing Ethical Biases and Jailbreak Vulnerabilities in AI Systems