LOADBench: A Large Open Analog Dataset and Benchmark for Machine Learning in IC Design
arXiv:2610.21591v1 Announce Type: new Abstract: Machine learning (ML) for analog IC design is constrained by limited open data and the absence of standardized evaluation: results reported in different papers are typically obtained on bespoke, often closed, datasets and with ad-hoc metrics, leaving comparisons across methods unreliable. We address both gaps with a large opensource dataset and a suite of standardized benchmarks. The dataset contains more than 100 operational amplifier (opamp) top…