>>940041717
If you provide a data set of 50+ pics of non-private persons, i'll consider most of the at least somewhat known people.
There are also other brewers and bakers who might also be willing to create a LoRA.

>>940041502
meta data viewer like
https://xypher7.github.io/lora-metadata-viewer/
are helpful to find out what you can use to trigger something related to the LoRA
1. The suggested prompts section contains the most frequent keywords
2.The data set definition contains the class prompt defining more or less what is to be expected to be genned in general. The claas prompt might be used as a fallback trigger if the regular triggerword is compromised or biased due to the checkpoint or other involved LoRA.
3. This is an overview of all used keywords to train the data set.
4. Clip skip controls which layer of the CLIP text encoder is used to generate image prompts—lower layers (e.g., skip = 2) tend to preserve more semantic meaning and improve facial detail. Choosing the right skip level can significantly affect how accurately the model responds to your prompts.