Nathan Helm-Burger comments on Google’s new 540 billion parameter language model

[Nathan Helm-Burger]

That’s already what TPUs do, basically