FineTune Writing Style from Dataset

Hey, since we can finetune gpt-3.5 i wanted to ask if the style of writing is also changing?
Like if i would finetune it on my speach would it write like I would?

That would be a very high hurdle and hard task.

First, you would not fine-tune on “your speech”. You would give example user questions and the types of responses the AI should give as training data.

Hundreds if not thousands of prompts “write an essay on soil amendments, in the style of lucameopou”, and then the affected language as you’d want the AI to produce.

Hi and welcome to the developer forum!

You can try leaving the user roll content blank and putting around 1000-1500 tokens worth of your transcribed speech as the assistant roll content, that is the standard way to train on open ended data, it should start to pick up on patterns in your speech and your style of writing, it does quite well with Shakespeare after a single book and is quite good after 2 or 3.