pytorch单GPU模型加载多GPU保存的参数

from collections import OrderedDict
state_dict = torch.load(para_path)
new_state_dict = OrderedDict()
for k, v in state_dict.items():
    name = k[7:] # remove `module.`
    new_state_dict[name] = v
net.load_state_dict(new_state_dict)