facebookresearch/chameleon

The generated image does not make any sense.

MaureenZOU opened this issue · 2 comments

I have tried the model with miniviewer, under the default setting, the generated image looks very strange, the text generation part seems reasonable. It would be really appreciated if anyone could help.
Screenshot 2024-07-21 at 9 29 03 PM

@MaureenZOU, they didn't release the image generation ability.
For image generation, you might refer to this: https://github.com/GAIR-NLP/anole?tab=readme-ov-file

Thanks Thao! Then this makes sense to me.