Text-to-Image
Diffusers
English
SVDQuant
FLUX.1-Krea-dev
FLUX.1
Diffusion
Quantization
ICLR2025

config.json and comfy_config.json missing files

#1
by vladulidlo - opened

Cannot load model in ComfyUI due to error:
OSError: Error no file named config.json found in directory /ComfyUI/models/diffusion_models/svdq-int4-flux.1-krea-dev

Edit: nevermind, I found out I only need to load the .safetensor file instead of whole folder like other Nunchaku svdq models.

It works fine in ComfyUI. You just need the ComfyUI-nunchaku extension.

It works fine in ComfyUI. You just need the ComfyUI-nunchaku extension.

I already have ComfyUI-nunchaku extension, but it complained about missing .json files in a folder with this model I selected to load as I always put in folder as other svdq models. I found out that using folder as model source does not work for new models with single file. Newer models point diretcly to single .safetensor and not a folder.

upgrade ComfyUI-nunchaku to the latest version (v0.3.3) and put the safetensors directly in models/diffusion_models.

upgrade ComfyUI-nunchaku to the latest version (v0.3.3) and put the safetensors directly in models/diffusion_models.

I already did, that was the confusing part, as Nunchaku was know to use model folder instead of model single .safetensor file.

Thank you for your response, I'm sure somebody will find this helpful.

Sign up or log in to comment