Revisiting LARS for Large Batch Training Generalization of Neural Networks

Open in new window