jsttlgdkycy

How to compute the FID of text-image models?

1

I'm curious about what dataset does FIDs shown in CLIP-FID curve computed with? And are the text-image FIDs reported in https://arxiv.org/abs/2112.10752 be computed use one of these four models? I...

About the Implementation of Classifier-Free Guidance

2

https://github.com/hpcaitech/Open-Sora/blob/2361353844da516bc711720c0cd82989e1467769/opensora/schedulers/iddpm/__init__.py#L91 Since there are four channels in the latent space, I wonder why only the first three channels are selected to apply CFG. Is this a typo or something else?...

How about the training setting of CelebA model

3

I am very interested in the training setting of the CelebA model since I retrained one but got a fid of only 4.5 with 1000 steps DDIM sampler. Could you...