How to use Pipeline Parallel to inference？

[uestcMeng]

Thanks to share your work! The demo you provided has only support single-gpu inferenec, and I find it seems that PP inference is not supported by diffsynth-studio. I want to know how to use the lora model when I load Qwen-Image-Edit using diffusers lib, like following code:

from diffusers import QwenImageEditPlusPipeline

pipeline = QwenImageEditPlusPipeline.from_pretrained("Qwen/Qwen-Image-Edit-2509", torch_dtype=torch.bfloat16)

How can I use this pipeline to load the lora model? Thanks in advance.

[witcherofresearch]

Hello, I have released the diffuser version of QwenStyle v1 in the readme, please check it. QwenStyle v1 is not very stable and we will soon release a better version QwenStyle v2.

[uestcMeng]

Thanks, I want to finetune QwenStyle v1 on my own dataset. I find that there is no training code in this repo, will you release the training code in the future? if so, thanks a lot.

[witcherofresearch]

I used Diffsynth-Studio for training.