---
title: ZeroGUI-AndroidLab-7B
canonical_url: "https://www.modelscope.cn/models/Shanghai_AI_Laboratory/ZeroGUI-AndroidLab-7B"
md_url: "https://www.modelscope.cn/models/Shanghai_AI_Laboratory/ZeroGUI-AndroidLab-7B.md"
repository: Shanghai_AI_Laboratory/ZeroGUI-AndroidLab-7B
last_updated: 2025-05-30
license: apache-2.0
pipeline_tag: image-text-to-text
tasks:
  - image-text-to-text
model_type:
  - qwen2_vl
architectures:
  - Qwen2VLForConditionalGeneration
base_model:
  - ByteDance-Seed/UI-TARS-7B-DPO
base_model_relation: finetune
parameters: 8.3B
tensor_type:
  - BF16
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
  - zh
inference_backends:
  - "deploy_task vlm/text/emb"
  - "lmdeploy 0.9.1"
  - "sglang 0.5.2"
  - "vllm 0.9.2"
downloads: 388
stars: 0
tags:
  - multimodal
  - gui
---

# ZeroGUI-AndroidLab-7B

> ZeroGUI-AndroidLab-7B - Shanghai_AI_Laboratory 在 ModelScope 开源的模型。ZeroGUI-AndroidLab-7B

Shanghai_AI_Laboratory/ZeroGUI-AndroidLab-7B 是 ModelScope 魔搭社区上的 8.3B 参数image-text-to-text模型，采用 apache-2.0 许可，基于 ByteDance-Seed/UI-TARS-7B-DPO 构建，可用 deploy_task vlm/text/emb、lmdeploy 0.9.1、sglang 0.5.2 部署。

- **Repository**: Shanghai_AI_Laboratory/ZeroGUI-AndroidLab-7B
- **License**: apache-2.0
- **Tasks**: image-text-to-text
- **Parameters**: 8.3B
- **Base model**: ByteDance-Seed/UI-TARS-7B-DPO
- **Inference backends**: deploy_task vlm/text/emb, lmdeploy 0.9.1, sglang 0.5.2, vllm 0.9.2
- **Tags**: multimodal, gui
- **Downloads**: 388
- **Stars**: 0
- **Last updated**: 2025-05-30

Source: https://www.modelscope.cn/models/Shanghai_AI_Laboratory/ZeroGUI-AndroidLab-7B

---

# ZeroGUI-AndroidLab-7B

[\[📜 Paper\]](https://arxiv.org/abs/2505.23762)
[\[📂 GitHub\]](https://github.com/OpenGVLab/ZeroGUI)

## Introduction

We propose **ZeroGUI**, a fully automated online reinforcement learning framework that enables GUI agents to train and adapt in interactive environments at zero human cost.

* **Automatic Task Generation:** Automatically proposes diverse, executable GUI tasks.

* **Automatic Reward Estimation:** Assigns binary task rewards based on trajectory screenshots and employs a voting mechanism to avoid hallucinated success.

* **Two-Stage Online RL:** Combines training on generated tasks and test-time adaptation to continually improve agent's performance.

![image/png](https://cdn-uploads.huggingface.co/production/uploads/637f347a52229c639211bee8/1vrET2pXAV8quJIqme0z3.png)

## Results

![image/png](https://cdn-uploads.huggingface.co/production/uploads/637f347a52229c639211bee8/FLbNDzy6aC9XV-VGAWYVA.png)

## Citation

If you find this work helpful in your research, please consider citing:

```bibtex
@article{yang2025zerogui,
  title={ZeroGUI: Automating Online GUI Learning at Zero Human Cost},
  author={Yang, Chenyu and Shiqian, Su and Liu, Shi and Dong, Xuan and Yu, Yue and Su, Weijie and Wang, Xuehui and Liu, Zhaoyang and Zhu, Jinguo and Li, Hao and Wang, Wenhai and Qiao, Yu and Zhu, Xizhou and Dai, Jifeng},
  journal={arXiv preprint arXiv:2505.23762},
  year={2025}
}
```