PPL
Jianing Zhu
Papers (7)
Your Agents Are Aging Too: Agent Lifespan Engineering for Deployed SystemsAgentHijack: Benchmarking Computer Use Agent Robustness to Common Environment CorruptionsRethinking How to Remember: Beyond Atomic Facts in Lifelong LLM Agent MemoryMicro-Defects Expose Macro-Fakes: Detecting AI-Generated Images via Local Distributional ShiftsWhen Is Rank-1 Steering Cheap? Geometry, Granularity, and Budgeted SearchAgentForesight: Online Auditing for Early Failure Prediction in Multi-Agent SystemsBelief Memory: Agent Memory Under Partial Observability