Bayesian Convolutionnal network with Dropout

#1

I’m trying to implement Bayesian Convolutionnal network with Dropout. I have defined two classes: the first class Net_do(nn.Module) (mcdo=Net_do()) for training data and another class Net(nn.Module) (net=Net()) for test data. During the training phase, I did Outputs=mcdo(inputs) and during the testing phase, I did Outputs=Net(inputs) I searched for help, but I can´t find a solution to my problem. I will leave the error here, I’m stuck I’m really gonna appreciate your help, Thank you

sa
sa831×421 23.4 KB

#2

So when we perform a model(...) call, it actually ends up calling model.forward(...). What you’re seeing here is that it’s calling that method (line 550) but apparently your Net_do model does not have such a method. Therefore your model instead inherits the forward method from its parent (nn.Module) which raises the NotImplementedError.

Does your Net_do have a forward method? If so, can you please post the class code for Net_do?

#3

Thanks for reply,
yes it has a forward method
1 2

#4

Is it Net_MCDO? The forward method is indented one too many times

1 Like
#5

Oh yes ! I just noticed
Thank’s a lot