世界を巻き戻し、反省を残す：長期LLMエージェントのためのロールバック誘導リフレクション
Rollback the World, Keep the Reflection: Rollback-Induced Reflection for Long-Horizon LLM Agents
長期タスクに取り組むLLMエージェントが失敗した際、環境状態を過去に巻き戻しつつ、捨てた軌跡から有用な知識を蒸留して引き継ぐことで、同じ過ちを繰り返さず効率的に回復する枠組みを提案。
詳しい要約
1. どんなもの？
2. 先行研究と比べてどこがすごい？
3. 技術・手法の肝は？
4. どうやって有効だと検証した？
5. 議論はある？
6. 次に読むべき論文は？
※ AIが要旨から生成した要約です。正確性は原文をご確認ください。
著者: Yi Yu, Liuyi Yao, Yaliang Li, Enshu Wang, Libing Wu
分類: cs.CL, cs.RO
原文アブストラクト
Large language model (LLM) agents increasingly tackle long-horizon tasks through multi-step environment interaction, yet a single erroneous action can alter subsequent states and observations, causing errors to compound over time. Existing methods either correct the context without repairing altered environment states or restore earlier states while discarding useful experience, making it difficult to both eliminate failure conditions and avoid repeating past mistakes. We argue that reliable recovery should instead be treated as a rollback-boundary control problem that jointly determines when to intervene, where to resume, and what information should survive recovery. Based on this view, we propose Rollback-Induced Reflection (RIR), a unified recovery framework that restores execution to a selected prior state while carrying forward reusable knowledge distilled from the abandoned trajectory to guide subsequent decisions. We further characterize recovery through a unified operator over rollback depth and retained memory, providing a general view of state restoration and knowledge retention. Experiments on three long-horizon benchmarks demonstrate that RIR consistently improves task performance across multiple LLM backbones, with structured reflection memory preserving useful experience and selective rollback enabling efficient recovery.