DAI-Lab/RivaGAN

Bad Accuracy

Opened this issue · 4 comments

LiTangCherry commented
  • RivaGAN version:
  • Python version: 3.6
  • Operating System: Ubuntu18.04

Hi, I trained the model with hollywood2 according to the guideline. But I only get an accuracy of 74% when there is no modification. Did I miss anything? Thank you!

Li

2459122312 commented

Hello，I have some problem when i try to use anaconda to repeat experiment.The Makefile can't be implemented successfully.
Could you please share your idea?Thank you!

xbtzpx commented
  • RivaGAN version:
  • Python version: 3.6
  • Operating System: Ubuntu18.04

Hi, I trained the model with hollywood2 according to the guideline. But I only get an accuracy of 74% when there is no modification. Did I miss anything? Thank you!

Li

Could you tell me if you have successfully reproduced the code? The author mentioned that both "Moments in Time" and "Hollywood2" datasets are quite large. Do you need to download both datasets? Could you provide your contact information so that I can reach out to you?

MakrisJr commented

Which branch is everyone using? the master one is missing things e.g., the data folder. I didn't manage to "git checkout stable"
or "make install" either

llbbcc commented

Maybe it is because Hollywood2 is too small.