Is there any differences between deepcopy() and lod_state_dict()?

vision
1 
1. Train A network.
2. B.load_state_dict(A.state_dict())

1. Train A network.
2. B = copy.deepcopy(A)

Is there any differences between deepcopy() and load_state_dict() ?
and this process means fine-tuning?