TheFerumn

Today i tried to use 1.5 model for something else and same error happened while trying normal generation. Reason was turned on token merging ratio 0.5. I am not sure...

> (Side note: If anyone knows better ARGs than what I'm using, please share) Well. I don't think you need --all-in-fp32 . Try fp16 and if everything works you should...

> > > (Side note: If anyone knows better ARGs than what I'm using, please share) > > > > > > Well. I don't think you need --all-in-fp32 ....

I have no idea what i am doing but after some debugging with ChatGPT it suggested me to edit autocast_mode.py into this: ``` import importlib from typing import Any, Optional...

Changing commands into --no-half --no-half-vae instead of --upcast-sampling fixes it. As far as i see its known issue.

Issue is still here but i can't fix it the same way as before. BTW: I am not sure what preprocessor_inpaint.py is for but inpainting works even with this error....

When i try to upscale image with img2img tab i am having this error ``` Traceback (most recent call last): File "C:\Stable Diffusion\new-forge\modules_forge\main_thread.py", line 30, in work self.result = self.func(*self.args,...

I found this issue https://github.com/lshqqytiger/stable-diffusion-webui-amdgpu/issues/503 and there is solution to do downgrade to fix it but is there any way to fix it in Forge ? @lshqqytiger ?

download LDM folder from this link: https://github.com/CompVis/stable-diffusion and drop it into E:\automatic_forge

> Of course, I spent 10 minutes trying to figure out exactly where to throw it, but I was able to. It really helped! > > However, my AnimateDiff interface...