让智能体根据问题难度动态调整配置，提升效率与准确率
Learning to Configure Agentic AI Systems
- 将配置选择建模为半马尔可夫决策过程，实现动态适配
- 推理准确率提升31.3%，工具使用准确率提升13.95%
- 适合需要高效自适应智能体系统的研发者参考
基于大语言模型的智能体系统配置涉及从大量组合空间中选择工作流、工具、令牌预算和提示词，当前通常依赖固定模板或手动调参，忽略查询难度差异，导致行为脆弱且计算浪费。为此，本文将配置问题形式化为半马尔可夫决策过程（SMDP），每个配置被视为一个时序扩展的选项，决定智能体如何处理查询，并提出ARC（Agentic Resource & Configuration learner）——一种轻量级分层策略，可动态选择针对特定查询的配置。在推理、工具使用及智能体基准测试中，ARC均显著优于同预算的工具增强型LLM：推理准确率提升31.3%，工具使用准确率提升13.95%，τ-Bench（Airline）通过率从9.0%提升至18.0%。结果表明，学习按查询定制配置是“一刀切”设计的强大替代方案。
原文摘要 · Abstract (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 templates or hand-tuned heuristics that apply the same configuration regardless of query difficulty, leading to brittle behavior and wasted compute. To address this, we formulate agent configuration as a semi-Markov decision process (SMDP) where each configuration acts as a temporally extended option that determines how an agent system processes a query, and introduce introduce ARC (Agentic Resource & Configuration learner), a lightweight hierarchical policy that dynamically selects query-specific agent configurations. Across reasoning, tool-use, and agentic benchmarks, ARC consistently improves over budget-matched tool-augmented LLMs, increasing average reasoning accuracy by 31.3%, tool-use accuracy by 13.95%, and doubling τ-Bench (Airline) Pass^1 success from 9.0% to 18.0%. These results demonstrate that learning per-query agent configurations is a powerful alternative to "one size fits all" designs.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。