txchmechanicus/hermes-agent-reasoning-traces
收藏资源简介:
Hermes Agent Reasoning Traces数据集是一个用于训练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。数据集有两个配置,分别对应不同的源模型:kimi(Moonshot AI Kimi-K2.5)和glm-5.1(ZhipuAI GLM-5.1-FP8)。数据集包含多轮对话、工具定义、任务类别和子类别等信息,对话消息使用ShareGPT格式。数据集涵盖了终端与编码、代理工具、仓库任务、浏览器自动化等多个类别,并提供了详细的统计信息和生成细节。
The Hermes Agent Reasoning Traces dataset is a collection 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, one per source model: kimi (Moonshot AI Kimi-K2.5) and glm-5.1 (ZhipuAI GLM-5.1-FP8). It includes multi-turn dialogues, tool definitions, task categories, and subcategories, with conversation messages in ShareGPT format. The dataset covers various categories such as Terminal & Coding, Agent Tools, Repository Tasks, and Browser Automation, and provides detailed statistics and generation details.




