Is this repo dead?

[CypherQube]

Is this repo dead?

[guaneec]

Sorry, I should have made the status of the project clearer. I'm not actively updating this repo for a couple reasons:

• I mainly made this for myself, but currently LoRA satisfies my needs and is easier to work with (native support, large community, tooling)
• The main reason why I made this is largely that I didn't like the licensing of the official repo. However, the original authors have made their implementation available in diffusers. They have also added an MIT License to the repo, making this motivation void.
• After using multiple implementations, I've grown to believe it's better to use CLI for training. The gradio GUI makes it harder to try out hyperparameter configurations. Also, the webui training UI is less informative and user friendly compared to TensorBoard or W&B.
• I'm kind of busy lately

If enough people still find value in this, I will probably update more. It will likely be a rewrite though, using diffusers a la kohya-sd-scripts-webui, since there's no reason to maintain a different storage format now.

[HeavyTony2]

If enough people still find value in this, I will probably update more. It will likely be a rewrite though, using diffusers a la kohya-sd-scripts-webui, since there's no reason to maintain a different storage format now.

That's sound an interresting plan. Good luck with your future projects, no matters if they concern custom-diffusion or not.