Support for SSD-1B

Hello kohya,

I've been making use of your wonderful repository for my projects, and I wanted to inquire if there's a possibility to introduce support for training the SSD-1 model. The inclusion of SSD-1 would benefit many users like me who are interested in leveraging its capabilities.
https://huggingface.co/segmind/SSD-1B

SSD-1 is known for its training speed faster than SDXL.

I understand that integrating a new model can be a significant undertaking, and I appreciate all the work that has gone into this repository so far. If there's any way the community can assist in this endeavor, please let us know.

Thank you for considering this request, and I look forward to any updates on this matter.

Best regards,
Alfredplpl

Start by reviewing the repository's existing model-training support and the SSD-1B model page linked in the issue. Identify the integration points and validation needed for SSD-1B training; done means the repository can train that model successfully.

Written by the indexing model from the issue text.