questions about the KID metric

[GunZ]

Hi,thank you for your great work!

After inferencing on VITON-HD dataset with your released model, I got KID_p 0.0015 (1.08 in your paper), and KID_u 0.0018 (1.60 in your paper). Why such a big difference？

Thanks for any advices.

[Alberto Baldrati]

Hi @Ganzhi-e
Thanks for your interest in our work!

As stated in #13, following previous works, we report the kid metric multiplied by a factor of 10^3 in our paper.

I hope this solves your issue,
Alberto

[GunZ]

Thank you very much !