PPL
Yuan Tian
Papers (7)
Enhancing Multi-Agent Communication through Attention Steering with Context RelevanceWhen Think-with-Image Meets Safety: What Determines Multimodal Jailbreak Robustness?HIDBench: Benchmarking Large Language Models for Host-Based Intrusion DetectionNo Attack Required: Semantic Fuzzing for Specification Violations in Agent SkillsThe Granularity Mismatch in Agent Security: Argument-Level Provenance Solves Enforcement and Isolates the LLM Reasoning BottleneckMuon-OGD: Muon-based Spectral Orthogonal Gradient Projection for LLM Continual LearningSkill Drift Is Contract Violation: Proactive Maintenance for LLM Agent Skill Libraries