pro_model.py 文件源码

def def_netF():
    vgg16 = M.vgg16()
    vgg16.load_state_dict(torch.load('vgg16-397923af.pth'))
    vgg16.features = nn.Sequential(
        *list(vgg16.features.children())[:9]
    )
    for param in vgg16.parameters():
        param.requires_grad = False
    return vgg16.features