The Basic Stack for Query Performance

Jerry: In the classic Google query sense are now also kind of a complex prompt. These days, in terms of practical implementations, the retriever is like embedding search over a vector database and then the synthesis model is a language model. Jerry: We think there's more complexity that needs to be explored beyond the current stack that's emerging today. That's why you see a lot of these stacks that for instance, combine like a vector database with your language model. It's because they're implementing the stack.