[Feature] Support for LoRA or full fine-tuning for the Wan-2.2 series.

[KZF-kzf]

Motivation

Are there any plans to add LoRA training support for the Wan 2.2 model series, such as the 5B TI2V, 14B I2V, and 14B-Animate models?

Related resources

No response

[Edenzzzz]

It's mostly about porting these models to FastVideo. We are a bit tied up recently, if anyone's interested, feel free to submit a PR and then we can help. Mainly need to add the configs, pipeline and model class

[Edenzzzz]

See https://hao-ai-lab.github.io/FastVideo/inference/add_pipeline.html

[SolitaryThinker]

Thanks for your interest, we are actually actively working on supporting all of our recipes, from finetuning to sparse distillation on the wan2.2 moe model. Stay tuned!

[KZF-kzf]

Thanks for your interest, we are actually actively working on supporting all of our recipes, from finetuning to sparse distillation on the wan2.2 moe model. Stay tuned!

Thanks！

[pinderpeng]

Thanks for your interest, we are actually actively working on supporting all of our recipes, from finetuning to sparse distillation on the wan2.2 moe model. Stay tuned!

Hello, is there a current release plan for FastWan 2.2-I2V-14B? If so, when is it expected? Thank you!

[vaskers5]

++++