让大模型通过元推理更聪明地选工具，提升数学任务表现
Meta-Reasoning Improves Tool Use in Large Language Models
- 先生成候选工具，再通过反思过程择优选择
- 在多个数学数据集上显著提升性能，包括分布内和分布外
- 适合需要精准工具调用的复杂推理任务
外部工具能帮助大语言模型完成原本难以胜任的任务。现有框架在测试时依赖简单的贪心解码选择工具，无论模型是否经过工具标注数据微调或使用上下文示例提示。相比之下，我们发现收集并比较合适的候选工具集合，具有更大潜力实现最优选择。本文提出 TECTON（Tool selection via meta-reasoning），一个两阶段系统：第一阶段通过自定义微调的语言建模头对任务进行推理并输出候选工具；第二阶段禁用该头，对之前的推理过程进行元推理（即反思）以做出最终决策。实验表明，TECTON在多个数学推理数据集上均带来显著提升，涵盖分布内与分布外场景。
原文摘要 · Abstract (English)
External tools help large language models succeed at tasks where they would otherwise typically fail. In existing frameworks, choosing tools at test time relies on naive greedy decoding, regardless of whether the model has been fine-tuned on tool-annotated data or prompted with in-context examples. In contrast, we find that gathering and choosing among a suitable set of candidate tools has greater potential to lead to an optimal selection. We present Tool selECTion via meta-reasONing (TECTON), a two-phase system that first reasons over a task and outputs candidate tools using a custom fine-tuned language modelling head. Then, with the custom head disabled, it meta-reasons (i.e., it reasons over the previous reasoning process) to make a final choice. We show that TECTON results in substantial gains--both in-distribution and out-of-distribution--on a range of math reasoning datasets.
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。