“We keep on to discover hyperscaling of AI models bringing about better general performance, with seemingly no end in sight,” a pair of Microsoft researchers wrote in October within a site article asserting the company’s significant Megatron-Turing NLG model, built-in co