Cacher is the code snippet organizer for pro developers

#scale推荐值：0.0001
#附加稀疏惩罚项到梯度下降过程中 x_{k+1} = x_{k} - \alpha_{k} * g^{k} 
#m.weight.data权值越小对梯度影响越小，证明对通道越不重要
def updateBN(scale, model):
  for m in model.modules():
    if isinstance(m, nn.BatchNorm2d):
      m.weight.grad.data.add_(scale*torch.sign(m.weight.data)) #L1 正则化