WorkflowBench

该数据集是一个大规模的微调数据集，包含了来自83个应用程序的1,503个API的106,763个样本，这些应用程序涵盖了28个类别。该数据集旨在提升大型语言模型的工作流编排能力。此外，数据集融合了现实世界的工作流和合成的查询，这为复杂工作流编排任务的微调提供了可能。该数据集的规模达到了106,763个样本，其任务重点在于工作流编排和自动化。

This is a large-scale fine-tuning dataset consisting of 106,763 samples of 1,503 APIs sourced from 83 applications across 28 categories. This dataset aims to enhance the workflow orchestration capabilities of large language models (LLMs). Furthermore, it integrates real-world workflows and synthetic queries, enabling fine-tuning for complex workflow orchestration tasks. With a total of 106,763 samples, this dataset focuses on workflow orchestration and automation.