H${}^2$R: Hierarchical Hindsight Reflection for Multi-Task LLM Agents

Shicheng Ye, Chao Yu, Kaiqiang Ke +2

Large language model (LLM)-based agents have shown strong potential in multi-task scenarios, owing to their ability to transfer knowledge across diverse tasks. However, existing ap…