does it support comfyui?
#1
by
t8star
- opened
I try it,but it seemly don't support comfyui
Yes, I can see a few issues from the ComfyUI error log. This model has QKV stored as separate tensors while in the ComfyUI-repackaged models they're joined together. They're also scaled independently it seems, so to make it work in ComfyUI it's not enough to just concatenate the tensors.
same
These work with Comfy (those that have comfyui in the filename)
https://huggingface.co/lightx2v/Hy1.5-Quantized-Models/tree/main
As well as these:
https://huggingface.co/Comfy-Org/HunyuanVideo_1.5_repackaged/tree/main/split_files
Where you can also find the 4step 480p t2v as a lora