ShareGPT-4o-Image

# 📚 ShareGPT-4o-Image

**ShareGPT-4o-Image** is a large-scale and high-quality image generation dataset, where all images are produced by **GPT-4o’s image generation capabilities**. This dataset is designed to align open multimodal models with GPT-4o’s strengths in visual content creation. It includes **45K text-to-image** and **46K text-and-image-to-image** samples, making it a useful resource for enhancing multimodal models in both image generation and editing tasks.

<!-- > ⚠️ **Statement**: **ShareGPT-4o-Image** is a distilled dataset from GPT-4o-Image, offering 4o-level data quality (_referring to data, not model capability_). **Janus-4o** is a fine-tuned version of Janus-Pro on this dataset, with added image editing support. Fine-tuning brings noticeable gains in image generation, but **Janus-4o still lags behind GPT-4o-Image in overall performance**. -->

## Dataset Overview

**ShareGPT-4o-Image** contains a total of **91K image generation samples** from GPT-4o, categorized as follows:

| Data Type | Number of Samples |
| :---------------------- | :----------------- |
| Text-to-Image | 45,717 |
| Text-and-Image-to-Image | 46,539 |
| **Total** | **92,256** |

## Image Files

The image data is packaged into `.tar` archives:

* `text_to_image_part_*.tar` contains images from the text-to-image set.
* `text_and_image_to_image_part_*.tar` contains images from the text-and-image-to-image set.

You can extract all images using the following script:

```bash
for f in *.tar; do
tar -xf "$f"
done
```

## Resources

* **GitHub**: [FreedomIntelligence/ShareGPT-4o-Image](https://github.com/FreedomIntelligence/ShareGPT-4o-Image)
* **Model**: [Janus-4o-7B on Hugging Face](https://huggingface.co/FreedomIntelligence/Janus-4o-7B)
* **Paper**: [arXiv:2506.18095](https://arxiv.org/abs/2506.18095)

## Citation

If you find our dataset helpful, please consider citing our work:

```
@misc{chen2025sharegpt4oimg,
title={ShareGPT-4o-Image: Aligning Multimodal Models with GPT-4o-Level Image Generation},
author={Junying Chen and Zhenyang Cai and Pengcheng Chen and Shunian Chen and Ke Ji and Xidong Wang and Yunjin Yang and Benyou Wang},
year={2025},
eprint={2506.18095},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2506.18095},
}
```

# 📚 ShareGPT-4o-Image
**ShareGPT-4o-Image** 是一款大规模高质量图像生成数据集，所有图像均由**GPT-4o**的图像生成能力生成。本数据集旨在使开源多模态模型对齐GPT-4o在视觉内容创作领域的优势。数据集包含约4.5万个文本到图像（text-to-image）样本与约4.6万个文本与图像到图像（text-and-image-to-image）样本，是提升多模态模型图像生成与编辑任务性能的优质资源。

> ⚠️ **声明**：**ShareGPT-4o-Image** 是从GPT-4o-Image中蒸馏得到的数据集，具备GPT-4o级别的数据质量（此处指数据本身，而非模型能力）。**Janus-4o** 是Janus-Pro在本数据集上的微调版本，新增了图像编辑支持。微调可在图像生成任务中带来显著性能增益，但**Janus-4o的整体性能仍落后于GPT-4o-Image**。

## 数据集概览
**ShareGPT-4o-Image** 总计包含来自GPT-4o的92,256个图像生成样本，分类如下：

| 数据类型 | 样本数量 |
| :---------------------- | :----------------- |
| 文本到图像（text-to-image） | 45,717 |
| 文本与图像到图像（text-and-image-to-image） | 46,539 |
| **总计** | **92,256** |

## 图像文件
图像数据被打包为`.tar`归档文件：
* `text_to_image_part_*.tar` 包含文本到图像（text-to-image）数据集对应的图像。
* `text_and_image_to_image_part_*.tar` 包含文本与图像到图像（text-and-image-to-image）数据集对应的图像。

您可通过以下脚本解压所有图像：
bash
for f in *.tar; do
tar -xf "$f"
done

## 资源链接
* **GitHub仓库**：[FreedomIntelligence/ShareGPT-4o-Image](https://github.com/FreedomIntelligence/ShareGPT-4o-Image)
* **模型权重**：[Hugging Face 上的 Janus-4o-7B](https://huggingface.co/FreedomIntelligence/Janus-4o-7B)
* **学术论文**：[arXiv:2506.18095](https://arxiv.org/abs/2506.18095)

## 引用方式
若本数据集对您的研究有所帮助，请引用以下论文：
bibtex
@misc{chen2025sharegpt4oimg,
title={ShareGPT-4o-Image: Aligning Multimodal Models with GPT-4o-Level Image Generation},
author={Junying Chen and Zhenyang Cai and Pengcheng Chen and Shunian Chen and Ke Ji and Xidong Wang and Yunjin Yang and Benyou Wang},
year={2025},
eprint={2506.18095},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2506.18095},
}