“We continue on to discover hyperscaling of AI models resulting in greater performance, with seemingly no end in sight,” a set of Microsoft scientists wrote in Oct within a weblog publish announcing the company’s significant Megatron-Turing NLG model, in-built collaboratio