Open hadaev8 opened 2 years ago
Hi,
Thank you for pointing this out. However, the difference in runtime should not be this large. Could you provide the code snippet you are using to compute the FID? I have not rigorously evaluated the new the PyTorch antialiasing flag yet. I will update here, and the library, if it generates comparable results.
-Gaurav
Here is colab example (with another dataset)
https://colab.research.google.com/drive/1T2nTSMi58LMkWaUPDqx3VRbCtUGsRKvH?usp=sharing
From 3 mins on a100 it went to 40 mins.
Also, Pytorch added antialiasing to interpolation, should it be used for speding up calculations?