让大模型推理与调用工具更对齐，提升决策可解释性。
R2IF: Aligning Reasoning with Decisions via Composite Rewards for Interpretable LLM Function Calling
- 用复合奖励机制融合格式、正确性与思维链有效性，指导模型决策。
- 在BFCL数据集上最高提升34.62%，思维链有效率达0.05。
- 适合需要高可信度工具调用的场景，如智能助手、自动化系统。
函数调用使大语言模型能对接外部工具，但现有基于强化学习的方法存在推理过程与工具调用决策不一致的问题。我们提出R2IF，一种面向可解释函数调用的推理感知强化学习框架，采用包含格式/正确性约束、思维链有效性奖励（CER）和规范-修改-价值奖励（SMV）的复合奖励，通过GRPO优化。在BFCL/ACEBench上的实验表明，R2IF相较于基线最高提升34.62%（Llama3.2-3B在BFCL上），平均思维链有效性为0.05，显著提升函数调用准确率与可解释性，助力可靠工具增强型大模型部署。
原文摘要 · Abstract (English)
Function calling empowers large language models (LLMs) to interface with external tools, yet existing RL-based approaches suffer from misalignment between reasoning processes and tool-call decisions. We propose R2IF, a reasoning-aware RL framework for interpretable function calling, adopting a composite reward integrating format/correctness constraints, Chain-of-Thought Effectiveness Reward (CER), and Specification-Modification-Value (SMV) reward, optimized via GRPO. Experiments on BFCL/ACEBench show R2IF outperforms baselines by up to 34.62% (Llama3.2-3B on BFCL) with positive Average CoT Effectiveness (0.05 for Llama3.2-3B), enhancing both function-calling accuracy and interpretability for reliable tool-augmented LLM deployment.
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。