Information
# 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.