---
license: mit
language:
- en
datasets:
- SamGiantEagle/ABC-1M
---
# AutoBrep: Autoregressive B-Rep Generation with Unified Topology and Geometry (SIGGRAPH Asia 2025)
[Xiang Xu*](https://samxuxiang.github.io/), [Pradeep Jayaraman*](https://www.research.autodesk.com/people/pradeep-kumar-jayaraman/), [Joseph Lambourne*](https://www.research.autodesk.com/people/joseph-george-lambourne/), [Yilin Liu](https://yilinliu77.github.io/), [Durvesh Malpure](https://www.research.autodesk.com/people/durvesh-malpure/), [Pete Meltzer](https://www.research.autodesk.com/people/pete-meltzer/)

[![arXiv](https://img.shields.io/badge/📃-arXiv%20-red.svg)](https://arxiv.org/abs/2512.03018) 
[![Code](https://img.shields.io/badge/🌐-Code%20-blue.svg)](https://github.com/AutodeskAILab/AutoBrep) 

![alt](autobrep_teaser.png)

AutoBrep is a decoder-only Transformer model that autoregressively generates B-Rep geometry and topology tokens.


### Citation
```
@inproceedings{xu2025autobrep,
  title={AutoBrep: Autoregressive B-Rep Generation with Unified Topology and Geometry},
  author={Xu, Xiang and Jayaraman, Pradeep and Lambourne, Joseph and Liu, Yilin and Malpure, Durvesh and Meltzer, Pete},
  booktitle={Proceedings of the SIGGRAPH Asia 2025 Conference Papers},
  pages={1--12},
  year={2025}
}
```