How to load full finetune for inference?

[BlackTea-c]

Feature request / 功能建议

Motivation / 动机

It seems like only lora inference example in README.MD

Your contribution / 您的贡献

test the full finetune(LTX-VIDEO,Cogxvideo)

[a-r-r-o-w]

Could you try something like this:

from diffusers import HunyuanVideoTransformer3DModel, HunyuanVideoPipeline

transformer = HunyuanVideoTransformer3DModel.from_pretrained("/path/to/checkpoint-3000/transformer", torch_dtype=torch.bfloat16)
pipe = HunyuanVideoPipeline.from_pretrained("hunyuanvideo-community/HunyuanVideo", transformer=transformer, torch_dtype=torch.float16)
pipe.to("cuda")

[BlackTea-c]

it works,thanks.