Correct `sample_size` of Stable Diffusion 2's unet to have correct width and height default
Since diffusers==0.9.0 the width and height is automatically inferred from the sample_size attribute of your unet's config. It seems like your diffusion model has the same architecture as Stable Diffusion 2 which means that when using this model, by default an image size of 768x768 should be generated. This in turn means the unet's sample size should be 96.
In order to suppress to update your configuration on the fly and to suppress the deprecation warning added in this PR: https://github.com/huggingface/diffusers/pull/1406/files#r1035703505 it is strongly recommended to merge this PR.
It's said on the readme that this model is trained on 1.4
Based on stable diffusion 1.4
It's said on the readme that this model is trained on 1.4
no, this model is trained on sdv2-base, the original model was trained on 1.4
@patrickvonplaten this is trained on sdv2-base, afaik this has 512px native resolution, not 768 px?
Hey @nousr ,
You're 100% right - sorry about this.