PPL
Hui Liu
Papers (5)
Variational Inference for Evidential Deep LearningFrom Seeing to Thinking: Decoupling Perception and Reasoning Improves Post-Training of Vision-Language ModelsReducing the Safety Tax in LLM Safety Alignment with On-Policy Self-DistillationModel-Agnostic Lifelong LLM Safety via Externalized Attack-Defense Co-EvolutionSelf-Distilled Trajectory-Aware Boltzmann Modeling: Bridging the Training-Inference Discrepancy in Diffusion Language Models