dr-tulu-sft-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 SFT Data

This dataset contains the SFT training data for DR Tulu, containing prompts and full trajectories including reasoning traces, tool calls, and answers with citations.
The source prompts are curated from [OpenScholar](https://huggingface.co/datasets/allenai/openscilm_queries), [Search Arena](https://huggingface.co/datasets/lmarena-ai/search-arena-24k), and short-form QA datasets inclduing [WebWalker-Silver](https://huggingface.co/datasets/callanwu/WebWalkerQA), [TaskCraft](https://huggingface.co/datasets/PersonalAILab/TaskCraft), [PopQA](https://huggingface.co/datasets/akariasai/PopQA) and [TyDiQA (English)](https://github.com/google-research-datasets/tydiqa).

**Important**: This does *not* contain the SFT subsets created using prompts from [MegaScicen](MegaScience/MegaScience), [HotpotQA](https://hotpotqa.github.io/), and ScholarQA. We will release those subsets shortly in a separate file.

## 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).

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

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

# DR Tulu 监督微调（Supervised Fine-Tuning, SFT）数据集

本数据集用于为DR Tulu提供监督微调训练数据，包含提示词（Prompt）、完整交互轨迹，涵盖推理过程、工具调用以及带引用的回答。

本数据集的源提示词精选自[OpenScholar](https://huggingface.co/datasets/allenai/openscilm_queries)、[Search Arena](https://huggingface.co/datasets/lmarena-ai/search-arena-24k)，以及包括[WebWalker-Silver](https://huggingface.co/datasets/callanwu/WebWalkerQA)、[TaskCraft](https://huggingface.co/datasets/PersonalAILab/TaskCraft)、[PopQA](https://huggingface.co/datasets/akariasai/PopQA)与[TyDiQA（英文）](https://github.com/google-research-datasets/tydiqa)在内的短格式问答（Question Answering, QA）数据集。

**重要说明**：本数据集未包含使用[MegaScicen](MegaScience/MegaScience)、[HotpotQA](https://hotpotqa.github.io/)与ScholarQA的提示词构建的监督微调子数据集。相关子数据集将尽快通过单独文件发布。

## 授权协议

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

搜集汇总

背景与挑战

背景概述

该数据集为DR Tulu提供监督微调训练数据，包含来自OpenScholar、Search Arena及多个QA数据集的提示和完整轨迹，如推理轨迹、工具调用和带引用的答案。它采用ODC-BY许可，专为研究和教育目的设计，但未涵盖MegaScicen等来源的子集。