compare-pytorch-and-torch-grads.py 文件源码

def getM(mods):
    for m in mods:
        if isinstance(m, legacy.nn.SpatialConvolution):
            m.gradWeight[m.gradWeight.ne(m.gradWeight)] = 0
            l.append(torch.norm(m.gradWeight))
        elif isinstance(m, legacy.nn.Linear):
            l.append(torch.norm(m.gradWeight))
        elif isinstance(m, legacy.nn.Concat) or \
             isinstance(m, legacy.nn.Sequential):
            getM(m.modules)