Pytorch 重量档案的形状有问题吗？

最近我研究了Retinaface的文章，并尝试用mobilenetv2替换主干。 但是，运行train.py时会出现错误：

    Traceback (most recent call last):
  File "E:/study2/retinaface-pytorch/train.py", line 132, in <module>
    model = RetinaFace(cfg=cfg, pretrained=pretrained).train()
  File "E:\study2\retinaface-pytorch\nets\retinaface.py", line 74, in __init__
    self.body = _utils.IntermediateLayerGetter(backbone, cfg['return_layers'])
  File "D:\toolsware\anaconda\envs\pytorch\lib\site-packages\torchvision\models\_utils.py", line 45, in __init__
    if not set(return_layers).issubset([name for name, _ in model.named_children()]):

AttributeError: 'NoneType' object has no attribute 'named_children'

---

什么是“某物”？@Rob sth=something，sry我拼写不正确