---
license: apache-2.0
datasets:
- TeleEmbodied/VISTA-UMI-5K
base_model:
- lerobot/pi05_base
tags:
- VLA
---

<h1 align="center">
  <font size="6">VISTA: Vision-Grounded and Physics-Validated Adaptation of UMI Data for VLA Training</font>
</h1>

<p align="center">
  <a href="https://tele-umi-vista.github.io"><img src="https://img.shields.io/badge/🏠_Project-Homepage-1f77b4" alt="Project"></a>
  <a href="https://arxiv.org/abs/2606.04708"><img src="https://img.shields.io/badge/arXiv-2606.04708-b31b1b" alt="arXiv"></a>
  <a href="https://github.com/TeleHuman/umi-vista"><img src="https://img.shields.io/badge/Code-VISTA-000000?logo=github" alt="Code"></a>
  <a href="https://huggingface.co/datasets/TeleEmbodied/VISTA-UMI-5K"><img src="https://img.shields.io/badge/🤗_Dataset-VISTA--UMI--5K-ffcc00" alt="VISTA-UMI-5K"></a>
  <a href="https://huggingface.co/datasets/TeleEmbodied/UMI-VQA-8M"><img src="https://img.shields.io/badge/🤗_Dataset-UMI--VQA--8M-4caf50" alt="UMI-VQA-8M"></a>
</p>

---

## 🔭 Overview
<div id="fig1" align="left">
  <img src="assets/overview.png" width="80%">
</div>

> 🚧 **Checkpoint coming soon in a few days**