RAP-260K

# RAP Instruct 260K

## Dataset details

[`rap_train_260k.json`](https://huggingface.co/datasets/Hoar012/RAP-260K/blob/main/rap_train_260k.json) is the full dataset used for training RAP-Phi3-V;
[`rap_train_210k.json`](https://huggingface.co/datasets/Hoar012/RAP-260K/blob/main/rap_train_210k.json) is a subset used for training RAP-LLaVA.

We also provide the script [`generate_negative.py`](https://huggingface.co/datasets/Hoar012/RAP-260K/blob/main/generate_negative.py) for generating additional negative samples.

**Paper Link:**
https://arxiv.org/pdf/2410.13360

**Project Page:**
https://hoar012.github.io/RAP-Project/

## Intended Use
**Primary intended uses:**
The primary use is research on the personalization of multimodal LLMs (MLLMs).

**Primary intended users:**
The primary intended users of the model are researchers and hobbyists in computer vision, natural language processing, machine learning, and artificial intelligence.

## Agreement
- The RAP dataset is available for non-commercial research purposes only, we do not own the rights to these images.
- You agree not to reproduce, duplicate, copy, sell, trade, resell or exploit for any commercial purposes, any portion of the images and any portion of derived data.

## Citation

```
@InProceedings{Hao_2025_CVPR,
author = {Hao, Haoran and Han, Jiaming and Li, Changsheng and Li, Yu-Feng and Yue, Xiangyu},
title = {RAP: Retrieval-Augmented Personalization for Multimodal Large Language Models},
booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR)},
month = {June},
year = {2025},
pages = {14538-14548}
}
```

# RAP Instruct 260K 数据集

## 数据集详情
[`rap_train_260k.json`](https://huggingface.co/datasets/Hoar012/RAP-260K/blob/main/rap_train_260k.json) 为用于训练 RAP-Phi3-V 的完整数据集；[`rap_train_210k.json`](https://huggingface.co/datasets/Hoar012/RAP-260K/blob/main/rap_train_210k.json) 为用于训练 RAP-LLaVA 的子集数据集。

我们还提供了脚本 [`generate_negative.py`](https://huggingface.co/datasets/Hoar012/RAP-260K/blob/main/generate_negative.py)，用于生成额外的负样本。

**论文链接：**
https://arxiv.org/pdf/2410.13360

**项目主页：**
https://hoar012.github.io/RAP-Project/

## 预期用途
**核心用途：**
本数据集主要用于多模态大语言模型（Multimodal Large Language Models，MLLMs）的个性化研究。

**核心适用人群：**
本数据集的核心适用人群为计算机视觉、自然语言处理、机器学习与人工智能领域的研究人员及爱好者。

## 使用协议
- RAP 数据集仅可用于非商业性研究用途，我们不拥有这些图像的版权。
- 您同意不得出于任何商业目的，对图像或衍生数据的任何部分进行复制、拷贝、售卖、交易、转售或利用。

## 引用格式

@InProceedings{Hao_2025_CVPR,
author = {Hao, Haoran and Han, Jiaming and Li, Changsheng and Li, Yu-Feng and Yue, Xiangyu},
title = {RAP: Retrieval-Augmented Personalization for Multimodal Large Language Models},
booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR)},
month = {June},
year = {2025},
pages = {14538-14548}
}