“We keep on to find out hyperscaling of AI models resulting in improved overall performance, with seemingly no conclude in sight,” a pair of Microsoft researchers wrote in October in the weblog publish announcing the company’s large Megatron-Turing NLG model, built-in coll