Test saved model accuracy on different validation sets


(sumit) #1

I have trained a model based on the resnet34 architecture for my problem facial expression recognition and saved it using

learn.save('224_all').

My problem is that I want to see the models accuracy on datasets compiled from different sources. Say, valid2 contains files from a different dataset. How do i go about doing that in fastai? I could load the saved model and call learn.fit again and see the validation accuracy but that won’t be very neat…