GaParmar / img2img-turbo

One-step image-to-image with Stable Diffusion turbo: sketch2image, day2night, and more
MIT License
1.55k stars 177 forks source link

'AutoencoderKL' object has no attribute 'add_adapter' #64

Open YingJGuo opened 3 months ago

YingJGuo commented 3 months ago

The AutoencoderKL has no attribute 'add_adapter' in diffusers 0.30 and 0.29.2. It works in diffusers==0.25.1. But I'm curious about why the latest version removed this attribute, and whether there will be anything wrong with using add_adapter to finetune the vae.

I am new to this field, sry if the question is stupid.

BeyondHeaven commented 3 months ago

Thanks for your comment! I have the same issue and find the solution from you.

HkDzl commented 1 month ago

I am also curious