Computer-Aided Design as Language
Authors: Yaroslav Ganin, Sergey Bartunov, Yujia Li, Ethan Keller, Stefano Saliceti
NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We collect a dataset of over 4.7M of carefully preprocessed parametric CAD sketches and use this dataset to validate the proposed generative models. To our knowledge, the experiments presented in this work significantly surpass the scale of those reported in the literature both in terms of the amount of training data and the model capacity. |
| Researcher Affiliation | Industry | Yaroslav Ganin1 Sergey Bartunov1 Yujia Li1 Ethan Keller2 Stefano Saliceti1 1Deep Mind 2Onshape |
| Pseudocode | Yes | Listing 1: Examples of object specifications. We represent objects using Protocol Buffers. Protocol Buffers allow us to easily write specifications for structured objects of varying complexity. (Includes structured code for message Line Entity, message Mirror Constraint, message Entity, message Constraint, message Object, message Sketch). |
| Open Source Code | No | The code is proprietary but may be open-sourced in the future. |
| Open Datasets | Yes | The dataset is available at https://bit.ly/3m9QHPd. |
| Dataset Splits | Yes | For our experiments we split the dataset randomly into 3 parts: 4,656,607 examples for the training set and 50,000 sketches for each the validation and the test set. |
| Hardware Specification | No | The paper states in its ethics review guidelines that it included details on compute resources, but these specific hardware details (e.g., exact GPU/CPU models, memory) are not found within the provided text of the paper. |
| Software Dependencies | No | The paper mentions software components like Transformer, Protocol Buffers, and Sentence Piece tokenizer, but it does not specify any version numbers for these or other software dependencies. |
| Experiment Setup | No | The paper does not provide specific hyperparameter values (e.g., learning rate, batch size, number of epochs) or detailed training configurations in the main text. It refers to Appendix E for details, but this appendix is not provided in the prompt. |