How to Setup a Tech Stack for Do-It

You can try different ways of telling the model on how it should operate. You can tell them to work on probabilities. And you can give it sort of an instruction that says, if you see a probability, if you don't feel that the probability of guessing the next token is above 95%, then just tell us. It might work well, but there is no guarantee, right? See if the model feels unsure, then obviously it's just a statistical parrot, right? Now we get to get therapists for models. Okay, okay. I'll stop. If you're adding context, can you add context from multiple documents at once? Yeah, so there's no limitation. Of course