Scaling MLPs: A Tale of Inductive Bias | Read Paper on Bytez