评测大模型何时该调用工具，提升决策能力。
When2Call: When (not) to Call Tools
- 设计新基准评估模型何时调用工具、追问或承认无法回答。
- 顶尖模型在该基准上表现不佳，说明决策能力仍有不足。
- 提出偏好优化训练法，效果优于传统微调，适合工具调用场景。
利用外部工具是现代语言模型扩展能力并集成到现有系统的关键。然而，现有基准主要关注工具调用的准确性——是否以正确参数调用了正确工具——而较少评估模型应何时（不）调用工具。为此，我们构建了新基准 When2Call，用于评估工具调用的决策能力：何时生成工具调用、何时提问追问，以及何时承认无法用所提供工具回答。我们发现，当前最先进的工具调用模型在 When2Call 上表现仍有显著提升空间，凸显该基准的重要性。我们还构建了 When2Call 的训练数据集，并利用基准的多选性质设计了一种偏好优化训练方法，其性能提升远超传统微调。相关代码、数据与评估脚本已开源：https://github.com/NVIDIA/When2Call。
原文摘要 · Abstract (English)
Leveraging external tools is a key feature for modern Language Models (LMs) to expand their capabilities and integrate them into existing systems. However, existing benchmarks primarily focus on the accuracy of tool calling -- whether the correct tool is called with the correct parameters -- and less on evaluating when LMs should (not) call tools. We develop a new benchmark, When2Call, which evaluates tool-calling decision-making: when to generate a tool call, when to ask follow-up questions and when to admit the question can't be answered with the tools provided. We find that state-of-the-art tool-calling LMs show significant room for improvement on When2Call, indicating the importance of this benchmark. We also develop a training set for When2Call and leverage the multiple-choice nature of the benchmark to develop a preference optimization training regime, which shows considerably more improvement than traditional fine-tuning. We release the benchmark and training data as well as evaluation scripts at https://github.com/NVIDIA/When2Call.
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。