SBER-MoVQGAN

SBER-MoVQGAN is a new SOTA model in the image reconstruction problem. This model is based on code from the VQGAN repository and modifications from the original MoVQGAN paper.

Code for using SBER-MoVQGAN you can obtain in our repo.

Models

The following table compares the 3 versions of the model SBER-MoVQGAN on the Imagenet dataset in terms of FID, SSIM and PSNR metrics. A more detailed description of the experiments and a comparison with other models can be found in the Habr post.

Model	Train steps	FID	SSIM	PSNR
f=8, SBER-MoVQGAN 67M	2M	0,96	0,7249	26,45
f=8, SBER-MoVQGAN 102M	2360k	0,78	0,7373	26,89
f=8, SBER-MoVQGAN 270M	1330k	0,69	0,7411	27,04