PPL
Dongrui Liu
Papers (7)
AgentDoG 1.5: A Lightweight and Scalable Alignment Framework for AI Agent Safety and SecurityPlant, Persist, Trigger: Sleeper Attack on Large Language Model AgentsPreference-aware Influence-function-based Data Selection Method for Efficient Fine-TuningAttention Hijacking: Response Manipulation Across Queries in Vision-Language ModelsTacoMAS: Test-Time Co-Evolution of Topology and Capability in LLM-based Multi-Agent SystemsAgentSlimming: Towards Efficient and Cost-Aware Multi-Agent SystemsOn the Blessing of Pre-training in Weak-to-Strong Generalization