A: Yes, we train a small model to write in the style of the provided samples. We actually fine tune the entire model around your writing (no LORA tricks if you are aware of that that means) which gives us a small model that can do one thing well - write in the style of the samples.

A few tips
* Remove any disclaimer text, rocs, artifacts, etc. from the samples
* Use at least 100K words and aim for 200K if you can or more
* Use samples from one human, or, at least use samples in a similar writing style.

What to expect:
* Writing style emulation
* No emulation of struturing style
* Longer book generation times when using a human model