I trained this network from this github.

The training went well, and returns nice results for new, unseen images.

On training, the loss changed (decreased), thus I must assume the weights changed as well.

On training, I saved a snapshot of the net every epoch.

When trying to run a validation set through each epoch's snapshot, I get the exact same results on every epoch.

How can this be possible? What's causing this?


The answer is I was too tired to see I was using the same model for validation every epoch, instead of that epoch's model.

Write good code, and this won't happen to you.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.