How is ‘Content Aware’ context added.

I had to turn this off on our use case, because after we fine-tuned it seemed using content aware was putting more of an emphasis on that vs what we specifically trained it for since the {Content} was in context. But I’m sure that’s case by case, although I’d love to know the weights on each.

Great observation. How are you finding the fine-tuned model?

My use case is basically just a digital health assistant that offers expanded knowledge on health articles. I’ve been able to get surprisingly great results with just dynamically adding the title and excerpt of the post (along with a custom field for additional context like related products or services) into the context. Keeps the prompt small and the bot still replies with it’s focussed scope of knowledge.