Add support for LoRA/Finetune/Control training of Wan 2.2 #423
Description
Feature request / 功能建议
The Wan Team recently released the Wan 2.2 Model with a MoE setup and better generation quality. Would love to see new methods to fine-tune it and/or create LoRAs for it. Happy to support/contribute for it!
Motivation / 动机
To enable LoRA/Control/SFT training for Wan 2.2 for the community.
Your contribution / 您的贡献
Edit: I'm currently working on this!