PPL
Ruixiang Tang
Papers (4)
TRACES: Proactive Safety Auditing for Multi-Turn LLM Agents via Trajectory-State ModelingMicro-Defects Expose Macro-Fakes: Detecting AI-Generated Images via Local Distributional ShiftsAgentForesight: Online Auditing for Early Failure Prediction in Multi-Agent SystemsA Single Layer to Explain Them All:Understanding Massive Activations in Large Language Models