i finally figured out how to stop my outputs from sounding like a robot by feeding the model samples of my own work to analyze. if u feed it enough examples of ur
tone and structure , the custom instructions actually start to work for ur
content workflow instead of against it.
it's basically just advanced pattern matching . does anyone else use a specific
style_guide.txt
file for their custom gpts?
found this here:
https://zapier.com/blog/train-chatgpt-to-write-like-you