Masala-CHAI
收藏资源简介:
Masala-CHAI是由纽约大学和康奈尔大学联合创建的SPICE网表数据集,旨在通过大型语言模型(LLMs)自动化生成模拟电路的SPICE网表。数据集包含约2100个从教科书中提取的电路图,涵盖了不同复杂度的模拟电路。创建过程中,采用了对象检测、深度Hough变换和提示调优等技术,确保网表的准确性。该数据集主要应用于模拟电路设计与验证领域,旨在解决传统手动生成网表的耗时和低效问题。
Masala-CHAI is a SPICE netlist dataset jointly developed by New York University and Cornell University. It aims to automate the generation of SPICE netlists for analog circuits using Large Language Models (LLMs). The dataset contains approximately 2,100 circuit diagrams extracted from textbooks, covering analog circuits of varying complexities. During its creation, techniques including object detection, deep Hough transform, and prompt tuning were adopted to ensure the accuracy of the netlists. This dataset is primarily applied in the field of analog circuit design and verification, aiming to address the time-consuming and inefficient issues of traditional manual netlist generation.




