Vision Foundation Models as Effective Visual Tokenizers for Autoregressive Image Generation

Open paper
Benchmark Model Rank Results
image-reconstruction-on-imagenetVFMTok#8FID: 0.89
image-generation-on-imagenet-256x256VFMTok-3B#103FID: 2.07
image-generation-on-imagenet-256x256VFMTok-XXL(SigLIP2)#114FID: 2.16
image-generation-on-imagenet-256x256VFMTok-XXL#119FID: 2.19
image-generation-on-imagenet-256x256VFMTok-L#146FID: 2.75
image-generation-on-imagenet-256x256VFMTok-B#159FID: 3.43

Rank counts only results with a code link.