Masala-CHAI: A Large-Scale SPICE Netlist Dataset for Analog Circuits by Harnessing AI
Masala-CHAI is the first fully automated framework using large language models (LLMs) to generate SPICE netlists for analog circuit design. By leveraging GPT-4’s multi-modal capabilities, it addresses the long-standing challenge of automating netlist generation from circuit schematics. Our three-step workflow—circuit labeling, prompt tuning, and netlist verification—enhances accuracy and efficiency in circuit design automation. Tested on 2,100 schematics of varying complexity, Masala-CHAI demonstrates significant improvements in netlist generation. We open-source this framework to drive further innovation in AI-powered circuit design.