arXiv 2409.009202025-07-25cs.LGcs.AIcs.CL

ToolACE：赢得 LLM 函数调用的得分点

ToolACE: Winning the Points of LLM Function Calling

Shanghai Jiao Tong University（上海交通大学）；University of Science and Technology of China（中国科学技术大学）；Huawei Noah’s Ark Lab（华为诺亚实验室）；Huawei Technologies Co., Ltd（华为技术有限公司）；Tsinghua University（清华大学）；The Chinese University of Hong Kong（香港中文大学）

Weiwen Liu, Xu Huang, Xingshan Zeng, Xinlong Hao, Shuai Yu, Dexun Li, Shuai Wang, Weinan Gan, Zhengying Liu, Yuanqing Yu, Zezhong Wang, Yuxian Wang, Wu Ning, Yu… 展开作者

Weiwen Liu, Xu Huang, Xingshan Zeng, Xinlong Hao, Shuai Yu, Dexun Li, Shuai Wang, Weinan Gan, Zhengying Liu, Yuanqing Yu, Zezhong Wang, Yuxian Wang, Wu Ning, Yutai Hou, Bin Wang, Chuhan Wu, Xinzhi Wang, Yong Liu, Yasheng Wang, Duyu Tang, Dandan Tu, Lifeng Shang, Xin Jiang, Ruiming Tang, Defu Lian, Qun Liu, Enhong Chen

备注：: 21 pages, 22 figures

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出自动智能体数据合成流水线 ToolACE，通过自演化 API 构建、多智能体对话生成和双层验证，训练出可匹敌 GPT-4 的 8B 函数调用模型。

摘要：

函数调用显著扩展了大型语言模型的应用边界，而高质量且多样化的训练数据对于释放这一能力至关重要。然而，真实的函数调用数据在收集和标注方面极具挑战，而现有流水线生成的合成数据往往缺乏覆盖度和准确性。本文提出 ToolACE，这是一条自动化智能体流水线，旨在生成准确、复杂且多样的工具学习数据。ToolACE 利用一种新颖的自演化合成过程，构建了包含 26,507 个多样化 API 的全面 API 池。在形式化思维过程的引导下，多个智能体之间的交互进一步生成对话。为确保数据准确性，我们实施了结合基于规则检查与基于模型检查的双层验证系统。我们证明，使用所合成数据训练的模型即使仅有 8B 参数，也能在 Berkeley Function-Calling Leaderboard 上达到最先进性能，与最新的 GPT-4 模型相匹敌。我们的模型和部分数据已公开发布于 https://huggingface.co/Team-ACE。

英文摘要：

Function calling significantly extends the application boundary of large language models, where high-quality and diverse training data is critical for unlocking this capability. However, real function-calling data is quite challenging to collect and annotate, while synthetic data generated by existing pipelines tends to lack coverage and accuracy. In this paper, we present ToolACE, an automatic agentic pipeline designed to generate accurate, complex, and diverse tool-learning data. ToolACE leverages a novel self-evolution synthesis process to curate a comprehensive API pool of 26,507 diverse APIs. Dialogs are further generated through the interplay among multiple agents, guided by a formalized thinking process. To ensure data accuracy, we implement a dual-layer verification system combining rule-based and model-based checks. We demonstrate that models trained on our synthesized data, even with only 8B parameters, achieve state-of-the-art performance on the Berkeley Function-Calling Leaderboard, rivaling the latest GPT-4 models. Our model and a subset of the data are publicly available at https://huggingface.co/Team-ACE.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。