This code is a mindspore implementation of SMU: smooth activation function for deep networks using smoothing maximum technique which is avaliable at https://github.com/iFe1er/SMU_pytorch.