Chat engine

Hi, can you elaborate your issue for me to understand better 🙂

Thank you for the response I saw query transformations under the different query engines like the hyde or multistep but i was wondering if there was anyway to breakdown the query for a chat engine?

Also I was wondering if there were any features for user personalization like saving user preferences?

@Logan M sorry to bother I was wondering if you had any thoughts about this

I mean, the default chat engine is an agent, which will break down queries for you. And of course, you can always call the LLM with a query/instructions directly if you wanted to create your own processing

Personalization should be achieved through the system prompt

Thank you so much for your response. I see I will start testing with the prompting to handle breaking down queries in the caht engine. Also sorry what do you mean about the system prompt?

index.as_chat_engine(...., system_prompt="Talk like a pirate in every response.")

The system prompt is just a message that gets inserted to the start of every LLM call. Most chat engines support it

Oh thank you and more what I mean about user preferences are like age, gender, etc. to help guide more tailored responses

I think you could put details like that in the system prompt right?

Ah I see that is true more my question is if there a tool right now for extracting those sort of details?

hmm not really. You mean extracting those through conversation? You'd have to do that yourself

Okay sounds good thank you again!