metadata
license: apache-2.0
datasets:
- kulibinai/cadevolve
base_model:
- Qwen/Qwen2-VL-2B-Instruct
CADEvolve: Realistic CAD Program Generation via Program Evolution
CADEvolve is a model trained to transform visual CAD inputs (e.g., multi-view renders) into 3D CAD models represented as runnable CadQuery Python code. It was trained on the CADEvolve dataset: kulibinai/cadevolve.
Code and training/inference utilities: GitHub.
Paper: arXiv:2602.16317.
Citation
@article{elistratov2026cadevolve,
title={CADEvolve: Creating Realistic CAD via Program Evolution},
author={Elistratov, Maksim and Barannikov, Marina and Ivanov, Gregory and Khrulkov, Valentin and Konushin, Anton and Kuznetsov, Andrey and Zhemchuzhnikov, Dmitrii},
journal={arXiv preprint arXiv:2602.16317},
year={2026}
}