CAD-LLM

Abstract

Parametric Computer-Aided Design (CAD) is the dominant paradigm for modern mechanical design. Training generative models to reason and generate parametric CAD can dramatically speed up design workflows. Pre trained foundation models have shown great success in natural language processing and computer vision. The cross-domain knowledge embedded in these models holds significant potential for understanding geometry and performing complex reasoning about design. In this work, we develop generative models for CAD by leveraging pre-trained language models and apply them to manipulate engineering sketches. Our results demonstrate that models pre-trained on natural language can be fine-tuned on engineering sketches and achieve remarkable performance in various CAD generation scenarios.