“We go on to find out hyperscaling of AI models resulting in much better general performance, with seemingly no finish in sight,” a set of Microsoft scientists wrote in Oct in a website article asserting the company’s enormous Megatron-Turing NLG model, in-built collaborat