Select different value of sensitivity from ROC curve to test a pre-trained model

Hi, I made the ROC curve on the validation_set of my dataset using a pre-trained model (“fpr, tpr, thresholds = roc_curve (…)”). Now I wanted to test this model on new images to be able to see the results of my work for this binary classification. At the moment I just used a “torch.max (output, 1)” to determine which class the input image belongs to. Is it possible to do this classification by modifying the choice according to the type of sensitivity I want?

You have two classes?
In this case you could use output[:, 0] >= thresh if you determined the threshold for class 0 (or change 0 to 1 if you used class 1.

Best regards