I'm not sure fine tuning helps with this kinda thing. Because the fine tuning mechanism involves sending the API a bunch of data like {"prompt": "<prompt text>

I'm not sure fine tuning helps with this kinda thing. Because the fine tuning mechanism involves sending the API a bunch of data like {"prompt": "<prompt text>", "completion": "<ideal generated text>"} So I'm not sure how you would feed it PG's essays in this mechanism