dreambooth: add support for .txt captions for instance images

[timh]

Not sure if instance captions are desired on the main dreambooth example, but this is what I've been using locally. FWIW, I haven't gotten great results with captions so far, which are mostly runs with 20-50 images of a person.

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint.

[devilismyfriend]

Dreambooth is meant to be used on a single token, with prior loss preservation

[patrickvonplaten]

This use case is a bit too exotic IMO to merge in to "main". We cannot maintain all kinds of extensions of dreambooth here sadly.

@pcuenca what do you think?

[pcuenca]

@patrickvonplaten I agree, I think it's a bit too specific. The training scripts are in my opinion meant to be examples, so people are free to extend and modify them to suit their needs, just as you did here @timh.

It's an interesting experiment, though, curious to know if you end up having good results, @timh :)

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.