学会配置智能体人工智能系统

Learning to Configure Agentic AI Systems

February 12, 2026

作者: Aditya Taparia, Som Sagar, Ransalu Senanayake

cs.AI

摘要

基於大型語言模型的智能體系統配置涉及從龐大的組合設計空間中選擇工作流程、工具、令牌預算及提示模板，目前通常採用固定的大型模板或人工調優的啟發式方法。這種方式會導致系統行為脆弱且產生不必要的計算消耗，因為無論輸入查詢難易與否，往往都套用相同的繁瑣配置。我們將智能體配置定義為按查詢決策的優化問題，並提出ARC（智能資源配置學習器），通過強化學習訓練輕量級分層策略來動態調整配置方案。在多個涵蓋推理任務和工具增強問答的基準測試中，學習得到的策略持續優於人工設計的強基準方法及其他對比方案，在實現任務準確率提升最高達25%的同時，有效降低了令牌消耗和運行時長。這些結果表明，針對單個查詢學習智能體配置是替代"一刀切"設計的有效範式。

English

Configuring LLM-based agent systems involves choosing workflows, tools, token budgets, and prompts from a large combinatorial design space, and is typically handled today by fixed large templates or hand-tuned heuristics. This leads to brittle behavior and unnecessary compute, since the same cumbersome configuration is often applied to both easy and hard input queries. We formulate agent configuration as a query-wise decision problem and introduce ARC (Agentic Resource & Configuration learner), which learns a light-weight hierarchical policy using reinforcement learning to dynamically tailor these configurations. Across multiple benchmarks spanning reasoning and tool-augmented question answering, the learned policy consistently outperforms strong hand-designed and other baselines, achieving up to 25% higher task accuracy while also reducing token and runtime costs. These results demonstrate that learning per-query agent configurations is a powerful alternative to "one size fits all" designs.