Tool-Calling-Dataset-UIGEN-X

# Tool Calling Dataset for UIGEN-X

This is a cleaned version of the [interstellarninja/hermes_reasoning_tool_use](https://huggingface.co/datasets/interstellarninja/hermes_reasoning_tool_use) dataset, specifically prepared for training with Axolotl.

## Dataset Description

- **51,004 training examples** of tool-use and reasoning conversations
- **Cleaned format** - Removed problematic `tools` column (tools are embedded in system messages)
- **Axolotl-compatible** - Works directly with `chat_template` type in Axolotl
- **Format**: Each example contains a `conversations` field with `from` and `value` keys

## Usage with Axolotl

```yaml
datasets:
- path: smirki/Tool-Calling-Dataset-UIGEN-X
type: chat_template
field_messages: conversations
message_property_mappings:
role: from
content: value
split: train
```

## Example Structure

```json
{
"conversations": [
{
"from": "system",
"value": "You are a deep thinking AI with tool-use capabilities..."
},
{
"from": "human",
"value": "User query..."
},
{
"from": "gpt",
"value": "Assistant response with tool calls..."
}
]
}
```

## Original Dataset

Based on [interstellarninja/hermes_reasoning_tool_use](https://huggingface.co/datasets/interstellarninja/hermes_reasoning_tool_use) with modifications for Axolotl compatibility.

## License

Apache 2.0 (inherited from original dataset)

# 面向UIGEN-X的工具调用数据集

本数据集为[interstellarninja/hermes_reasoning_tool_use](https://huggingface.co/datasets/interstellarninja/hermes_reasoning_tool_use)的清洗版本，专为适配Axolotl训练流程而制作。

## 数据集说明

- **51,004条工具调用与推理对话训练样本**
- **清洗后格式**：移除了存在问题的`tools`字段（工具调用指令已嵌入系统消息中）
- **Axolotl兼容**：可直接在Axolotl中使用`chat_template`（对话模板）类型
- **数据格式**：每条样本均包含`conversations`字段，该字段带有`from`与`value`两个键

## Axolotl使用方法

yaml
datasets:
- path: smirki/Tool-Calling-Dataset-UIGEN-X
type: chat_template
field_messages: conversations
message_property_mappings:
role: from
content: value
split: train

## 样本结构

json
{
"conversations": [
{
"from": "system",
"value": "You are a deep thinking AI with tool-use capabilities..."
},
{
"from": "human",
"value": "User query..."
},
{
"from": "gpt",
"value": "Assistant response with tool calls..."
}
]
}

## 原始数据集

本数据集基于[interstellarninja/hermes_reasoning_tool_use](https://huggingface.co/datasets/interstellarninja/hermes_reasoning_tool_use)制作，针对Axolotl兼容性进行了修改。

## 许可证

Apache 2.0（继承自原始数据集）

搜集汇总

背景与挑战

背景概述

该数据集是'interstellarninja/hermes_reasoning_tool_use'的清理版本，专门为Axolotl训练设计，包含51,004个工具使用和推理对话的训练示例。其特点是采用清理后的格式（工具嵌入系统消息中）并兼容Axolotl的'chat_template'，便于直接用于模型训练。