Is there a way to convert the existing model to gguf?
Hello,
I would like to convert my Z-Image Turbo checkpoint to GGUF.
Do you know of any simple conversion tools or Python scripts that can be used on Windows?
If there are any recommended tools, scripts, or reference resources, I would appreciate your guidance.
Thank you.
Hello,
I would like to convert my Z-Image Turbo checkpoint to GGUF.
Do you know of any simple conversion tools or Python scripts that can be used on Windows?If there are any recommended tools, scripts, or reference resources, I would appreciate your guidance.
Thank you.
Most diffusion ggufs are made with this repo: https://github.com/city96/ComfyUI-GGUF with some tweaks
Hello,
I would like to convert my Z-Image Turbo checkpoint to GGUF.
Do you know of any simple conversion tools or Python scripts that can be used on Windows?If there are any recommended tools, scripts, or reference resources, I would appreciate your guidance.
Thank you.Most diffusion ggufs are made with this repo: https://github.com/city96/ComfyUI-GGUF with some tweaks
I’m already aware of that.
While Flux can be converted to GGUF, Qwen and Z-Image have different architectures, which makes conversion difficult or currently unsupported.
Are there any existing conversion tools or Python scripts available for this purpose?
My goal is to convert Qwen and Z-Image models into GGUF format.
try sd.cpp
https://github.com/leejet/stable-diffusion.cpp/releases
To create an image, you can use CompyUI...
I need to convert it, but I can't confirm the method.
I'm trying to get advice from people who make gguf files.
As a model maker, bf16 is heavier than I thought, so it's better quality than fp8, and I'm thinking of making it a gguf file as light as fp8.
GGUF can be created but it depends on where you want to use it.
https://github.com/leejet/stable-diffusion.cpp/issues/1112
https://github.com/leejet/stable-diffusion.cpp/blob/master/docs/z_image.md
GGUF can be created but it depends on where you want to use it.
https://github.com/leejet/stable-diffusion.cpp/issues/1112
https://github.com/leejet/stable-diffusion.cpp/blob/master/docs/z_image.md
I’m aware that, due to the Diffusers format, models with different internal structures are currently difficult to convert.
I was just wondering whether conversion might still be possible in some way, so I wanted to ask.
In the case of Flux, conversion works because its structure is compatible, but for Qwen or Z-Image, it doesn’t seem to be possible yet.
I also tried converting them using sd.cpp, but I ran into the same errors, so it didn’t work either.
I guess I’ll need to look into it a bit more. Thank you.