I’m wondering if I could use GPT-3 for my usecase.
My goal is to create a “support chatbot” based on both:
- A set of docs
- History of previously answered questions
From what I’ve gathered I could either fine tune my own model with my set of docs or I could use the /answers enpoint with the set of previously answered questions. Is there a way to combine these two? Can I for example give the previously answered questions as documents for the fine tuning?
The docs and the questions span a variety of topics.