[2601.16771] AutoRegressive Generation with B-rep Holistic Token Sequence Representation

Computer Science > Computer Vision and Pattern Recognition arXiv:2601.16771 (cs) [Submitted on 23 Jan 2026 (v1), last revised 30 Mar 2026 (this version, v2)] Title:AutoRegressive Generation with B-rep Holistic Token Sequence Representation Authors:Jiahao Li, Yunpeng Bai, Yongkang Dai, Hao Guo, Hongping Gan, Yilei Shi View a PDF of the paper titled AutoRegressive Generation with B-rep Holistic Token Sequence Representation, by Jiahao Li and 5 other authors View PDF HTML (experimental) Abstract:Previous representation and generation approaches for the B-rep relied on graph-based representations that disentangle geometric and topological features through decoupled computational pipelines, thereby precluding the application of sequence-based generative frameworks, such as transformer architectures that have demonstrated remarkable performance. In this paper, we propose BrepARG, the first attempt to encode B-rep's geometry and topology into a holistic token sequence representation, enabling sequence-based B-rep generation with an autoregressive architecture. Specifically, BrepARG encodes B-rep into 3 types of tokens: geometry and position tokens representing geometric features, and face index tokens representing topology. Then the holistic token sequence is constructed hierarchically, starting with constructing the geometry blocks (i.e., faces and edges) using the above tokens, followed by geometry block sequencing. Finally, we assemble the holistic sequence representation for ...