I am just interested what kind of LoRA training works best for you (cloud/local), what you're using and, if you like, share your thoughts and configs.

I wanted to use kohya_ss, but get all types of errors, which led me back to ai-toolkit, although I don't really have a lot of experience with good training parameters for characters.

There are some tutorials, but they could be outdated. So that's why I am asking. Thanks!