Mamba-CAD: State Space Model for 3D Computer-Aided Design Generative Modeling

Authors: Xueyang Li, Yunzhong Lou, Yu Song, Xiangdong Zhou

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Comprehensive experiments are conducted to demonstrate the effectiveness of our model under various evaluation metrics, especially in the generation length of valid parametric CAD sequences.
Researcher Affiliation Academia Xueyang Li, Yunzhong Lou, Yu Song, Xiangdong Zhou B School of Computer Science, Fudan University, Shanghai, China EMAIL, EMAIL, EMAIL, EMAIL
Pseudocode Yes The pseudo code of Mamba-CAD can be found in the uploaded Technical Appendix .
Open Source Code No The paper states it will make its *dataset* publicly available in the future, but there is no explicit statement or link indicating that the *code* for the methodology described in this paper is currently available.
Open Datasets No Finally, total 77,078 CAD models described as parametric CAD sequences are grouped into a new dataset that will be released publicly to promote future research on the parametric CAD modeling. ...To train Mamba-CAD, we further construct a new dataset of parametric CAD sequences that is more complex than the existing benchmark dataset of the same type and will make it publicly available in the future.
Dataset Splits Yes Finally, total 77,078 CAD models are collected as represented in parametric CAD sequences, in which 61,662 CAD models for training, 7,707 models for validation and 7,709 models for testing.
Hardware Specification Yes The pre-training is conducted on one NVIDIA RTX 3090 GPU with the batch size of 32 under 10 epochs in about 5 hours.
Software Dependencies No The paper mentions several software-related concepts and frameworks (e.g., Transformer, Mamba, PyTorch implicitly through loss functions), but does not provide specific version numbers for any libraries, programming languages, or other software components.
Experiment Setup Yes The pre-training is conducted on one NVIDIA RTX 3090 GPU with the batch size of 32 under 10 epochs in about 5 hours. The initial learning rate is set to 0.001 with warmup (He et al. 2016) and gradient clipping of 1.0 is applied in back-propagation. ...train a 1-D Latent GAN(Chen and Zhang 2019) with 200,000 epochs and batch size of 256 in about 3 hours.