Mixed mode prompt engineering?

I’m curious how GPT-3 / OpenAI processes and handles this scenario:
I build a file with both prompt as blank and prompt with text. When the file is indexed, do the records with prompt text get indexed similarly to the ones that are empty?

2 Likes

I have a file of a few hundred records and in that same file I have both prompt and no prompt records. I’m trying to make sure that I’m tuning things in a single file.

1 Like

Yes! So, does classification-like index similarly to generation-like? Do prompt questions access both types and use them both for generation of results? Or do I need to create a generation like file and a classification file so that my prompts in fine tuning my question with also participate in my language engine generating results?

I’ve been seeing great results with this process but I’m not sure if I’m getting lucky or if I’m using it correctly for my needs. :slight_smile:

1 Like