NansException with Automatic1111

#1
by f1yingbanana - opened

I've downloaded both 16fp and 32fp models from beta 1 and 2, but all of them throw NansException when attempting to generate any image. WD 1.4 works fine.

modules.devices.NansException: A tensor with all NaNs was produced in Unet. This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check.

Adding the command line arg --no-half does not help, and adding --disable-nan-check generates a black image.

Adding the command line arg --no-half does not help, and adding --disable-nan-check generates a black image.

Hello. Same. Do You fix it somehow?

did you download the html?

I get something similar when I use stable diffusion 2.1

add the YAML file

Sign up or log in to comment