Notices by Sophie Huiberts (sophiehuiberts@mathstodon.xyz)

remember alphachip, a 2021 google paper that claimed that their RL-based methods outperformed the state-of-the-art methods for chip macro placement?

according to this new meta-analysis, their results dont replicate and are full of errors. there's also been allegations of fraud 👀

https://cacm.acm.org/research/reevaluating-googles-reinforcement-learning-for-ic-macro-placement

@julesh i admit, i hadn't thought about that interpretation yet. personally i was thinking about matters of which questions are interesting or worth studying and why. how someone answers those questions is core to their mathematical practice and is profoundly different between people, places and decades

i made a graph of which discord friends have met IRL, and it really illustrates the concept of the 'giant component' from random graph theory