PPL
Yifan Wu
Papers (5)
Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order OptimizationOut of Sight, Not Out of Mind: Unveiling Latent Attack in Latent-based Multi-Agent SystemsDiagEval: Trajectory-Conditioned Diagnosis for Reliable Software Evaluation with GUI AgentsAgentic Recommender System with Hierarchical Belief-State MemoryToken Economics for LLM Agents: A Dual-View Study from Computing and Economics