CADEvolve: Creating Realistic CAD via Program Evolution
Paper • 2602.16317 • Published • 30
Error code: JobManagerCrashedError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
CADEvolve is a synthetic dataset for CAD reverse engineering built around executable CadQuery programs. It is produced by the CADEvolve pipeline, which evolves a small set of primitives into 8k complex parametric generators, and then expands them into a large corpus of CAD programs (≈1.3M scripts) exercising a broad CadQuery operation set.
This dataset repo contains three top-level folders:
CADEvolve-G/
parametric_generators.json: the library of parametric generator programs (the evolved generator set).embeddings/: precomputed embeddings for retrieval / indexing (optional utility).CADEvolve-P/
Three subfolders with .py CAD programs:
ABC-P/CADEvolve-P-core/ShapeNet-P/CADEvolve-C/
Three subfolders with .py CAD programs:
ABC-C/CADEvolve-C-core/ShapeNet-C/@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}
}