Let’s say I have 10 Assistants created via API, each one with a different base instruction. What do I have to do if I want to “train” all of these assistants?
Suppose I want every assistant to write as if they are talking to a child and I already have instantiated these assistants. I am confused about how to achieve this: Do I modify the assistant instruction via POST https://api.openai.com/v1/assistants/{assistant_id} for every assistant or do I need to do fine-tuning?
It would be cool if we had a “base assistant” which is trained to talk like a child, and every other assistant was derived from it. Does something like that already exist?
Are you trying to get 10 assistants to generate responses in parallel and independently ? Or are you trying to get 10 assistants to interact with each other?
Sorry if it was not clear. I have 10 assistants that have specific purposes. (Like writing a blog, acting as a dietitian etc.) What I want is to train them all, at once. Like making sure they are writing as if they are explaining to a child.
I was wondering what is the best approach to this. I could modify the instructions on each of them, 1 by 1. Or I could try to fine-tune. Neither of these seem efficient though.
What I would love to have a base assistant, that already is “trained” to write as if it is explaining to a child. And other assistants, such as dietitian, could inherit that.
So “asst_diet” is an assistant that has dietitian instructions, “asst_blogger” is an assistant that has some blogger instructions. And I want to train them both. I was wondering if there is a system like this available. These assistants will act independently.