Missing key(s) in state_dict: “module.resnet50.conv1.weight“

在使用训练好的模型验证时，报以下错误：

    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for DataParallel:
	Missing key(s) in state_dict: "module.resnet50.conv1.weight", "module.resnet50.bn1.weight", "module.resnet50.bn1.bias", "module.resnet50.bn1.running_mean", "module.resnet50.bn1.running_var", "module.resnet50.block1.0.conv1.weight", "module.resnet50.block1.0.bn1.weight", "module.resnet50.block1.0.bn1.bias", "module.resnet50.block1.0.bn1.running_mean", "module.resnet50.block1.0.bn1.running_var", "module.resnet50.block1.0.conv2.weight", "module.resnet50.block1.0.bn2.weight", "module.resnet50.block1.0.bn2.bias", "module.resnet50.block1.0.bn2.running_mean",……

Unexpected key(s) in state_dict: "resnet50.conv1.weight", "resnet50.bn1.weight", "resnet50.bn1.bias", "resnet50.bn1.running_mean", "resnet50.bn1.running_var", "resnet50.bn1.num_batches_tracked", "resnet50.block1.0.conv1.weight", "resnet50.block1.0.bn1.weight", "resnet50.block1.0.bn1.bias", "resnet50.block1.0.bn1.running_mean", "resnet50.block1.0.bn1.running_var", "resnet50.block1.0.bn1.num_batches_tracked", "resnet50.block1.0.conv2.weight", "resnet50.block1.0.bn2.weight", "resnet50.block1.0.bn2.bias", "resnet50.block1.0.bn2.running_mean", "resnet50.block1.0.bn2.running_var", "resnet50.block1.0.bn2.num_batches_tracked", "resnet50.block1.0.conv3.weight", "resnet50.block1.0.bn3.weight", "resnet50.block1.0.bn3.bias", "resnet50.block1.0.bn3.running_mean", "resnet50.block1.0.bn3.running_var", "resnet50.block1.0.bn3.num_batches_tracked", "resnet50.block1.0.shortcut.0.weight", 

踩坑：

百度时查到此种解决方案，虽然不报错了，也可以运行了。
但是！！！！！！准确率只有0.01，哭了。明明训练时显示有90%多的准确率呀。

model.load_state_dict(state_dict，False)

解决办法

对于此种错误的原因，
网上大部分都说是训练的使用使用了nn.DataParallel ,而验证时未使用导致的。
但是我的代码里明明用到了多GPU的。

后来发现时加载顺序问题，应该先加载模型再调用nn.DataParallel！！！！！！

model.load_state_dict(torch.load(save_path))
model = nn.DataParallel(model)