Agent-Ark/Toucan-1.5M

Toucan-1.5M是一个由真实世界MCP环境合成的超过150万轨迹的最大的完全合成工具-代理数据集，用于推动代理型大型语言模型中的工具使用。该数据集包含多个配置，每个配置都有不同的特征和训练集大小，能够生成多样化的、真实的、具有挑战性的任务，这些任务需要使用多个工具，并涉及多轮、多转、顺序和并行的工具调用。

Toucan-1.5M is the largest fully synthetic tool-agent dataset to date, designed to advance tool use in agentic LLMs. It comprises over 1.5 million trajectories synthesized from 495 real-world Model Context Protocols (MCPs) spanning 2,000+ tools. The dataset includes multiple configurations, each with different features and training set sizes, capable of generating diverse, realistic, and challenging tasks that require the use of multiple tools, involving multi-round, multi-turn, sequential, and parallel tool calls.