can I calculate loss on specific output value with pytorch binary classifier

Question

I am building a binary classifier.

I want to calculate loss on predicted value 1 only, ignoring 0.

        criterion = nn.BCELoss()
        preds = mod