knowurknottty/hermes-agent-reasoning-traces

Hermes Agent Reasoning Traces是一个用于训练AI代理的多轮工具调用轨迹数据集。每个样本都是真实的代理对话，包含逐步推理（`<think>`块）和实际工具执行结果。数据集包含两个配置：kimi（来自Moonshot AI Kimi-K2.5模型）和glm-5.1（来自ZhipuAI GLM-5.1-FP8模型）。数据集涵盖了多种任务类别，如终端与编码、代理工具、存储库任务、浏览器自动化等。数据集使用ShareGPT格式存储对话消息，并包含详细的统计信息和类别分类。

Hermes Agent Reasoning Traces is a dataset of multi-turn tool-calling trajectories for training AI agents. Each sample is a real agent conversation with step-by-step reasoning (`<think>` blocks) and actual tool execution results. The dataset has two configs: kimi (from Moonshot AI Kimi-K2.5 model) and glm-5.1 (from ZhipuAI GLM-5.1-FP8 model). It covers various task categories such as Terminal & Coding, Agent Tools, Repository Tasks, Browser Automation, etc. The dataset uses ShareGPT format for conversation messages and includes detailed statistics and category classifications.