dr-tulu-rl-data

> [!NOTE]
> For full information, go check out the Dr Tulu paper [here](https://arxiv.org/abs/2511.19399).

<img src="https://huggingface.co/rl-research/DR-Tulu-SFT-8B/resolve/main/dr_tulu_logo.png" alt="Figure 1" width="500"/>

# DR Tulu RL Data

This dataset contains the RL training data for DR Tulu, containing prompts and search-based rubrics generated from OpenScholar and SearchArena prompts, with rubrics generated using GPT-4.1.

**Important**: This does *not* contain the RaR datasets we use in final RL training, but only the OpenScholar and SearchArena subsets. For the RaR data, we use data from:

1. [anisha2102/RaR-Science-20k-o3-mini](https://huggingface.co/datasets/anisha2102/RaR-Science-20k-o3-mini)
2. [anisha2102/RaR-Medicine-20k-o3-mini](https://huggingface.co/datasets/anisha2102/RaR-Medicine-20k-o3-mini)

We sample the first 3000 samples from RaR-Science, and the first 1000 samples from RaR-Medicine.
We will supply code for converting this data into a setup suitable for our training code in our [github](https://allenai.org/papers/drtulu).

## License

This dataset is licensed under ODC-BY. It is intended for research and educational use in accordance with [Ai2's Responsible Use Guidelines](https://allenai.org/responsible-use).

> **注意**：如需获取完整信息，请查阅Dr Tulu相关论文[点击此处](https://arxiv.org/abs/2511.19399)。

> 

# DR Tulu 强化学习数据集

本数据集为DR Tulu的强化学习（Reinforcement Learning）训练数据，包含源自OpenScholar与SearchArena提示词生成的提示内容与基于搜索的评分标准，其中评分标准由GPT-4.1生成。

**重要说明**：本数据集**不包含**我们在最终强化学习训练中使用的RaR数据集，仅包含OpenScholar与SearchArena子集。如需获取RaR数据集，请使用以下来源的数据：
1. [anisha2102/RaR-Science-20k-o3-mini](https://huggingface.co/datasets/anisha2102/RaR-Science-20k-o3-mini)
2. [anisha2102/RaR-Medicine-20k-o3-mini](https://huggingface.co/datasets/anisha2102/RaR-Medicine-20k-o3-mini)

我们从RaR-Science数据集中选取前3000条样本，从RaR-Medicine数据集中选取前1000条样本。
我们将在[官方GitHub仓库](https://allenai.org/papers/drtulu)中提供将本数据集转换为适配我们训练代码格式的代码。

## 许可协议

本数据集采用ODC-BY许可协议进行授权，仅可用于研究与教育用途，并需遵守[Ai2负责任使用指南](https://allenai.org/responsible-use)。