---
title: sharegpt4v-images
canonical_url: "https://www.modelscope.cn/datasets/hjh0119/sharegpt4v-images"
md_url: "https://www.modelscope.cn/datasets/hjh0119/sharegpt4v-images.md"
repository: hjh0119/sharegpt4v-images
last_updated: 2024-05-23
license: "Apache License 2.0"
storage_size: "97 GB"
downloads: 28083
stars: 2
---

# sharegpt4v-images

> sharegpt4v-images - hjh0119 在 ModelScope 开源的数据集。image datasets for sharegpt4v

hjh0119/sharegpt4v-images 是 ModelScope 魔搭社区上的数据集，存储大小 97 GB，采用 Apache License 2.0 许可。

- **Repository**: hjh0119/sharegpt4v-images
- **License**: Apache License 2.0
- **Storage size**: 97 GB
- **Downloads**: 28083
- **Stars**: 2
- **Last updated**: 2024-05-23

Source: https://www.modelscope.cn/datasets/hjh0119/sharegpt4v-images

---

image datasets for [sharegpt4v](https://www.modelscope.cn/datasets/AI-ModelScope/ShareGPT4V/summary)

as claimed in [data.md](https://github.com/InternLM/InternLM-XComposer/blob/main/projects/ShareGPT4V/docs/Data.md)
, you should organize the data as follows
```none
ShareGPT4V
├── ...
├── data
│   ├── llava
│   │   ├── llava_pretrain
│   │   │   ├── images
│   ├── coco
│   │   ├── train2017
│   ├── sam
│   │   ├── images
│   ├── gqa
│   │   ├── images
│   ├── ocr_vqa
│   │   ├── images
│   ├── textvqa
│   │   ├── train_images
│   ├── vg
│   │   ├── VG_100K
│   │   ├── VG_100K_2
│   ├── sharegpt4v
│   │   ├── share-captioner_coco_lcs_sam_1246k_1107.json
│   │   ├── sharegpt4v_instruct_gpt4-vision_cap100k.json
│   │   ├── sharegpt4v_mix665k_cap23k_coco-ap9k_lcs3k_sam9k_div2k.json
│   ├── share_textvqa
│   │   ├── images
│   ├── web-celebrity
│   │   ├── images
│   ├── web-landmark
│   │   ├── images
│   ├── wikiart
│   │   ├── images
├── ...
```

#### 下载方法 
:modelscope-code[]{type="git"}