---
dataset_info:
  features:
  - name: prompt
    dtype: string
  - name: thinking
    dtype: string
  splits:
  - name: split_0
    num_bytes: 5808433053
    num_examples: 200000
  download_size: 2480658009
  dataset_size: 5808433053
configs:
- config_name: default
  data_files:
  - split: split_0
    path: data/split_0-*
license: cc-by-4.0
task_categories:
- reinforcement-learning
- question-answering
- text-generation
language:
- en
tags:
- code
- llama4
- nvidia
- opencode
- reasoning
- nvidiadataset
size_categories:
- 100K<n<1M
---



<p align="center">
  <img src="https://huggingface.co/datasets/NuclearAi/HyperThink-Max-200K/resolve/main/A_logo_for_%22HyperThink%22_is_displayed_on_a_white_ba.png" alt="HyperThink Logo" width="600"/>
</p>

# 🔮 HyperThink

**HyperThink** is a premium, best-in-class dataset series capturing deep reasoning interactions between users and an advanced Reasoning AI system. Designed for training and evaluating next-gen language models on complex multi-step tasks, the dataset spans a wide range of prompts and guided thinking outputs.

---

## 🚀 Dataset Tiers

HyperThink is available in **three expertly curated versions**, allowing flexible scaling based on compute resources and training goals:

| Version | Samples | Description                                                                  |
| ------: | ------: | ---------------------------------------------------------------------------- |
|  `mini` |     50K | Lightweight, perfect for experimentation, fine-tuning, or rapid prototyping. |
|  `midi` |    100K | Balanced mid-scale version with diversity and depth.                         |
|   `max` |    200K | Full-power, production-grade dataset with rich reasoning paths.              |
|  `Code` |    200K | Full-power, production-grade dataset with rich in code reasoning .           |

---

## 📦 Dataset Structure

Each sample in the dataset contains:

- **prompt**: The original user input .
- **thinking**: The reasoning chain or intermediate thought process .
- **answer**: The final output or answer following the reasoning .

``` json
{
  "prompt": "Why do objects float in water?",
  "thinking": "Objects float when their density is lower than the density of 
}
```

---
## 📚 Use Cases

- **Chain-of-thought prompting**
- **Reasoning-based fine-tuning**
- **Instruction-following training**
- **Causal and deductive reasoning model evaluations**

---
## 🌐 Citation

If you use **HyperThink** in your work, please cite it:

``` bibtext
@dataset{nuclearai2025hyperthink,
  title     = {HyperThink: A Reasoning-Centric Prompt-Response Dataset},
  author    = {NuclearAi},
  year      = {2025},
  url       = {https://huggingface.co/datasets/NuclearAi/HyperThink-X-Nvidia-Opencode-Resoning-200K}
}
```

---

**Jay Shree Ram 🚩🚩**