RuntimeError for webcam.py

[hieuristic]

I have trained my model with 91% accuracy and saved in the save folder, but when I ran the command

$ python webcam.py --resume save/model_best.pth.tar,

the terminal return an error:

RuntimeError: Error(s) in loading state_dict for DataParallel: Unexpected key(s) in state_dict: "module.att.ChannelGate.mlp.0.weight", "module.att.ChannelGate.mlp.0.bias", "module.att.ChannelGate.mlp.2.weight", "module.att.ChannelGate.mlp.2.bias", "module.att.SpatialGate.spatial.0.weight", "module.att.SpatialGate.spatial.0.bias".,

I was wondering if the error raise from the code itself for my model.

[01smito01]

By default webcam.py expects the loaded model to be:

• Resnet 18
• Pretrained
• No attention module

Loading a model following those parameters cleared the dictionary errors for me - albeit replacing them with a new error:
File "webcam.py", line 56, in inference
pred = softmax(output[0].data).numpy()
TypeError: can't convert cuda:0 device type tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.

Which I resolved by replacing line 56 with
pred = torch.Tensor.cpu(softmax(output[0].data)).numpy()

[liuyiming199721]

问一下这个是什么注意力