cvlab-stonybrook/PathLDM

FID scores reported in the paper on training set?

Closed this issue · 3 comments

Hi authors,

I had one question regarding the FID reported in the paper, is it on training reports as mentioned in the Limitation, it wasn't clear just by reading the experiments section and Limitations section, so I wanted to be sure.

Yes, we obtained an FID ~ 10 on the test split

sorry, did you mean train split?

Oh ok, nvm, I see that train split is ~7. and you are saying on test split you got ~10.