You're really God~~~ Thanks!

#1
by Jason233 - opened

You're really God~~~ Thanks!

UnetLoaderGGUFAdvanced
Unexpected architecture type in GGUF file, expected one of flux, sd1, sdxl, t5encoder but got 'wan'

I had the same error message. I had to update the custom_node ComfyUI-GGUF. After that it worked.

UnetLoaderGGUFAdvanced
Unexpected architecture type in GGUF file, expected one of flux, sd1, sdxl, t5encoder but got 'wan'

Two things I've seen work for that.
I had the same problem with the t2v K_M, except it was telling me 'got pig'.

One person suggested forcing update of the Comfyui-GGUF node. He had the 'got wan' result. Updating the node worked for him. I'm screwed and I can't even see a version number at the moment. I git cloned and still not working for me. Also looks identical for changes for Wan in the recent changed py files so I don't know what's going on there.

But the other person I spoke to tried the K_S version instead of K_M and got it going. t2v is now working for me with a K_S but about 6x slower than the normal model. About to try the i2v.

So try updating the ComfyUI-GGUF node perhaps and if desperate like me run a K_S quant. Best I've got for you. Good luck.

edit : Oh I'm running other people's workflows and running Zluda, so the K_S might not be much help

it worked as you mention by updating theComfyUI-GGUF repo Thanks...

I could say nothing but thx. Best wishes.

Updating the ComfyUI-GGUF doesn't fix this issue.

Still encountering this issue myself. Updated within ComfyUI, git cloned the repo, tried everything but still getting 'got pig' error

Sign up or log in to comment