About the dataset for Car encoder training

[qilimk]

Hi @omertov,

Your research is so awesome and thanks for sharing your code. I am using the Car encoder, but I realize that the training dataset for the car encoder is Stanford Cars, not the LSUN Cars which is used for CAR StyleGAN-2 training. Is there any particular reason to switch to Stanford Cars? I see that all other encoders still use the same dataset as the original StyleGAN.

Thanks!