arXiv：cs.AI· Aditya Taparia, Som Sagar, Ransalu Senanayake·· 7 小时前AI 评分43
让 AI 智能体学会按需配置：ARC 用强化学习动态选择工作流与工具
Learning to Configure Agentic AI Systems
AI 导读
研究者提出 ARC（Agentic Resource & Configuration learner），一种分层策略，将智能体配置建模为半马尔可夫决策过程（SMDP），按查询难度动态选择工作流、工具、token 预算与提示词。
正文
Abstract: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 {\tau}-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.
| Comments: | 22 pages, 12 figures |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2602.11574 [cs.AI] |
| (or arXiv:2602.11574v5 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2602.11574 arXiv-issued DOI via DataCite |
Submission history
From: Aditya Taparia [view email]
[v1]
Thu, 12 Feb 2026 04:45:44 UTC (2,638 KB)
[v2]
Wed, 20 May 2026 05:20:13 UTC (3,352 KB)
[v3]
Thu, 21 May 2026 03:54:07 UTC (3,343 KB)
[v4]
Sat, 5 Sep 2026 23:22:21 UTC (3,343 KB)
[v5]
Tue, 6 Oct 2026 05:32:17 UTC (3,343 KB)
来源：arXiv：cs.AI · arxiv.org