About nan during evaluation

[potato-kitty]

Hello PNPinversion team, thank you for your wonderful works and your contribution for making these codes public, as well as the dataset. I am now using your evaluation framework and dataset for experiment toward some other methods, they are amazingly convenience for use and works really efficiently. However there is one thing I am not very sure about. There are some 'nan' occurred during evaluation, and I understand the reason for this. The point is that how should I deal with them? Or, more specifically, how you deal with them during your experiment, of which results are shown in your paper? Should they be just ignore, or set to 0 during calculating the final average results?
Really hope for your answer. Once again, thanks a lot for sharing your impressive work.

[gaspardbd]

Hello, that is a really good answer, I was wondering as well.