SHI-Labs/Versatile-Diffusion

Zero-Shot Fid

NieShenRuc opened this issue · 2 comments

Thanks for your excellent work! But I do not find zero-shot fid on MSCOCO of text-to-image, could you please provide more information?

Hi, we are still working on that part. On the other hand, we think MSCOCO FID is a temp solution to evaluate text-to-image because diffusion models run low on these scores, which doesn't match the real performance.

Ok, Thaks for your replay!