发表机构
Sony AI; EPFL; Sony Semiconductor Solutions; TU Munich（索尼人工智能; 洛桑联邦理工学院; 索尼半导体解决方案公司; 慕尼黑工业大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究提出混合合成框架LLM-SPICEMixer，将遗传网表生成与IGEL结合，在Iris分类任务中，相比无LLM引导的遗传框架，提升了多项性能指标。
AI 中文摘要
模拟电路拓扑合成仍具挑战性，因为有用的设计仅占据组合搜索空间的极小部分，且微小的结构变化会引发高度非线性的性能变化。进化算法颇具吸引力，因为它仅通过黑盒评估即可优化离散电路拓扑，但通常需要大量SPICE仿真，且可能过早收敛。本文提出LLM-SPICEMixer，这是一种混合合成框架，它将遗传网表生成与基于大型语言模型的提议算子IGEL（Inspiration-Guided Evolution with LLMs，即灵感引导进化）相结合。在搜索过程中，IGEL会用精英集中性能优异的电路提示大型语言模型，并指示其生成新的SPICE网表，该网表随后由SPICE评估，并采用与传统遗传算子相同的奖励机制进行选择。因此，大型语言模型提供结构化的拓扑提议，而仿真仍是判定的依据。我们在一项具有挑战性的基准任务上评估LLM-SPICEMixer：合成实现Iris分类判别函数的晶体管级电路。与无大型语言模型引导的遗传框架相比，LLM-SPICEMixer将中位数最终训练奖励提高了8.4%，中位数验证选择的测试奖励提高了8.8%。最佳验证选择的电路在标称tt角处达到93.3%的测试准确率，在17个工艺、电压和温度角上的平均测试准确率为85.9%。
英文摘要
Analog circuit topology synthesis remains challenging because useful designs occupy a tiny fraction of a combinatorial search space, and small structural changes can induce highly nonlinear changes in behavior. Evolutionary algorithms are attractive because they can optimize over discrete circuit topologies using only black-box evaluations, but they often require many SPICE simulations and may converge prematurely. We introduce LLM-SPICEMixer, a hybrid synthesis framework that augments genetic netlist generation with IGEL (Inspiration-Guided Evolution with LLMs), an LLM-based proposal operator. During search, IGEL prompts an LLM with high-performing circuits from the elite set and instructs it to generate a new SPICE netlist, which is then evaluated by SPICE and selected using the same reward mechanism as conventional genetic operators. Thus, the LLM contributes structured topology proposals while simulation remains the source of truth. We evaluate LLM-SPICEMixer on a challenging benchmark task: synthesizing transistor-level circuits that implement a discriminant function for Iris classification. Compared with the genetic framework without LLM guidance, LLM-SPICEMixer improves the median final training reward by 8.4% and the median validation-selected test reward by 8.8%. The best validation-selected circuit achieves 93.3% test accuracy at the nominal tt corner and 85.9% average test accuracy across 17 process, voltage, and temperature corners.
CommentsAccepted for MLCAD 2026, Extended Version
