Comparison with Transfusion

[NickGao96]

Transfusion seems to be also about AR + Diffusion Multi Modality Model (https://huggingface.co/papers/2408.11039). Are you using similar techniques? Is there any major difference?

[Sierkinhane]

Hi, sorry for the late reply. You can find two distinct differences i) representations for multimodal understanding, clip-vit and magvitv2 (ours) vs vae (transfusion); ii) representations for generation, magvitv2 (ours) vs vae (transfusion). More details can be found in our paper. Btw, welcome to star our repository.