TypeError: torchvision.transforms.transforms.RandomCrop is not a Module subclass

transforms = torch.nn.Sequential(
    transforms.RandomCrop(224),
    transforms.Normalize((0.25516997,0.25296235,0.24626788), (0.28230251,0.23966205,0.24667897))
    )
scripted_transforms = torch.jit.script(transforms)

338 if not isinstance(module, Module) and module is not None:
339 raise TypeError("{} is not a Module subclass".format(
–> 340 torch.typename(module)))
341 elif not isinstance(name, torch._six.string_classes):
342 raise TypeError(“module name should be a string. Got {}”.format(
TypeError: torchvision.transforms.transforms.RandomCrop is not a Module subclass

报错原因，pytorch1.6版本不能使用Sequential连接transforms，改为使用compose或者升级到1.9就好了。