reasoning-base-20k

# Dataset Card for Reasoning Base 20k

## Dataset Details

### Dataset Description

This dataset is designed to train a reasoning model. That can think through complex problems before providing a response, similar to how a human would. The dataset includes a wide range of problems from various domains (science, coding, math, etc.), each with a detailed chain of thought (COT) and the correct answer. The goal is to enable the model to learn and refine its reasoning process, recognize and correct mistakes, and provide high-quality, detailed responses. This dataset is currently in-progress.

- **Curated by:** [Nishith Jain](https://huggingface.co/KingNish)
- **Language(s) (NLP):** English
- **License:** Apache-2.0
- **Chat Template**: RChatML
```python
{%- for message in messages %}
{{- '<|im_start|>' + message['role'] + '\n' }}
{{- message['content'] + eos_token + '\n' }}
{%- endfor %}
{%- if add_generation_prompt %}
{{- '<|im_start|>assistant\n' }}
{%- endif %}
{%- if add_reasoning_prompt %}
{{- '<|im_start|>reasoning\n' }}
{%- endif %}
```

## Uses

### Direct Use

- **Model Training**: Train reasoning models to improve their ability to think through complex problems.
- **Research**: Study the effectiveness of different reasoning strategies and techniques.

## Dataset Structure

### Data Fields

- **user**: The user's query or problem statement.
- **assistant**: The correct answer to the problem.
- **reasoning**: A detailed, step-by-step reasoning process that explains how to arrive at the correct answer.
- **template**: A preapplied RChatML chat template.

# 推理基础20k数据集卡片
## 数据集详情
### 数据集描述
本数据集旨在训练推理模型，使其能够像人类一样，在生成回复前对复杂问题进行逐步思考推导。本数据集涵盖科学、编程、数学等多个领域的多样化问题，每个问题均附带详细的思维链（Chain of Thought, COT）与正确答案。本数据集的目标是帮助模型学习并优化其推理流程，识别并修正错误，最终生成高质量且详尽的回复。当前本数据集仍处于开发完善阶段。
- **整理方：** [Nishith Jain](https://huggingface.co/KingNish)
- **自然语言处理所用语言：** 英语
- **许可证：** Apache-2.0
- **对话模板：** RChatML
python
{%- for message in messages %}
{{- '<|im_start|>' + message['role'] + '
' }}
{{- message['content'] + eos_token + '
' }}
{%- endfor %}
{%- if add_generation_prompt %}
{{- '<|im_start|>assistant
' }}
{%- endif %}
{%- if add_reasoning_prompt %}
{{- '<|im_start|>reasoning
' }}
{%- endif %}

## 用途
### 直接用途
- **模型训练：** 训练推理模型，提升其处理复杂问题的思考推导能力。
- **研究：** 探究不同推理策略与技术的有效性。
## 数据集结构
### 数据字段
- **user：** 用户的查询语句或问题描述
- **assistant：** 问题的正确答案
- **reasoning：** 用于阐释正确答案推导逻辑的详尽分步推理过程
- **template：** 已预应用的RChatML对话模板。