ParentFull threadWanderPanda·For inference I agree! But training requires centralized gradient stepsView on HN