Error with Comfy converted checkpoint
I do realize this was literally just uploaded, but I did convert the model and tried it with both a Z-Image workflow and a Chroma workflow and got this error
UNETLoader
Error(s) in loading state_dict for NextDiT:
size mismatch for x_embedder.weight: copying a param with shape torch.Size([3840, 3072]) from checkpoint, the shape in current model is torch.Size([3840, 64]).
No VAE. It is pixel space.
SAME - a sample JSON workflow to show it working would be extremely helpful
Make sure you update your Comfy first and that you're at least on this commit https://github.com/Comfy-Org/ComfyUI/commit/9ebee0a2179b361a24c20838c1848d7988320636
Here's the workflow, just to show that it's working. For the sampler/steps/cfg/other params, I don't know which ones are best.
Make sure you update your Comfy first and that you're at least on this commit https://github.com/Comfy-Org/ComfyUI/commit/9ebee0a2179b361a24c20838c1848d7988320636
Here's the workflow, just to show that it's working. For the sampler/steps/cfg/other params, I don't know which ones are best.
Thank you! I will try this out! It seems all I need to do is change vae to pixel_space. Appreciate it! I know the model is not fully trained yet but I still want to check it out since I am a big fan of Chroma.
unusable