Honglin Mu

large language model (5) hierarchical planning (1) constrained generation (1) in-context learning (1) text generation (1) model safety (1) responsible ai (1) machine reading comprehension (1) model alignment (1) adversarial attack (1) black-box attack (1) language model (1) jailbreak attack (1) citation prediction (1) agent system (1) long-context understanding (1) long-form text (1) context compression (1) dynamic evaluation (1) behavioral prior (1)