Hey folks, just getting my feet wet. I

I mean, you could use postgres as a vectordb

Ah good call. I guess I would have thought "traditional" dbs would have been less efficient than "vector first" versions?

postgres/pgvector have made some pretty big strides, it seems fairly performant, plus with all the pluses of a normal sql db

Gotcha, I'll give it a go! One more for ya if you don't mind.

Started with this: https://blog.streamlit.io/build-a-chatbot-with-custom-data-sources-powered-by-llamaindex/ but I only ever get the top two k

How can I return top 10 for example?

also works for as_query_engine() (the kwarg that is)

I couldn't find that option in as chat engine but will give it a go.

Thank you much!

Yea, we abuse kwargs a bit here. It ends up being a kwarg to the underlying vector index retriever 🙂

Ah gotcha. As far as kwargs go is there a global list of what I can use where?

not really 😅 But, it might make more sense if you dig into the docs a bit

as_query_engine() is really a combination of a retriever, node postprocessors (if specified) and a response synthesizer

Then, a chat engine is a wrapper that calls either a query engine or just the retriever (depending on the mode)

So you can maybe tell, as_chat_engine() accepts all kwargs to the

• underlying chat engine (condense question in this case)
• vector store index retreiver
• response synthesizer

I've been diving deep into the docs but unwrapping what dictionary/kwargs I can pass, I haven't quite wrapped my head around. Just gotta stare it at more I guess. I'm more of a brute force dev where I learn by breaking stuff

Thanks again for your help! Much appreciated 👍

Yea for sure! Best of luck!

Omg I swore I did that last time but apparently not. Worked perfect!

One more if you don't mind @Logan M ...how do I integrate RetrieverQueryEngine in this example if I want to feed it into a index.as_chat_engine?

https://gpt-index.readthedocs.io/en/latest/examples/retrievers/reciprocal_rerank_fusion.html#use-in-a-query-engine

Maybe said another way, how do I turn that into a chat with history?

index.as_chat_engine() is a quick wrapper around a bunch of classes.

Once you have a query engine, you can initialize the class yourself.

For example
https://docs.llamaindex.ai/en/stable/examples/agent/openai_agent_with_query_engine.html

Bingo! Thanks a bunch!