wan2.1 training

[zyang6]

Please Does the warehouse support wan2.1 training?

[asomoza]

Hi, diffusers has very limited training for the video models, only for cogvideo I think. The video training is done in another repo called finetrainers which is maintained by some of the same people.

[zyang6]

Oh, sorry！why is it limited? I see there is a training for graph-generated texts.And is this repository a sub-repository of diffusers?

[asomoza]

sorry I missed this.

why is it limited?

because the training scripts of diffusers are mostly for learning and they are almost the very basic setup for training with not that much complexity and features so people can modify them to suit their needs, that's why they are inside the examples subdirectory. Maybe @a-r-r-o-w can answer this question better but I think it is mostly because the video training scripts aren't that basic.

is this repository a sub-repository of diffusers?

No, as I told you, that repo is maintained by a diffusers maintainer but it's not a priority for the diffusers team for now, maybe in the future this will change.