Hi! Is there an illustration about how

The indexes are pretty straight forward.

Vector index retrieves a top k

Summary index retrieves everything, no matter what

Is there any flowchart type of illustration how it works? Now I'm reading this section: https://docs.llamaindex.ai/en/stable/examples/response_synthesizers/long_context_test/

and it is very interesting to me. Especially the evaluation part.

But I don't really understand how each of those response_mode s are implemented to make hypotheses about why some response modes work in some cases and don't work in other cases. Having some illustrations would help me to choose a response mode based on my use case.

Theres no flow chart. I mean, theres two parts to the logic, how you retrieve text (i.e. retrieval) and how you use that text to make a response (response synthesis)

So retrieval, it litterally just takes everything you've put into it, no top k, no filtering

For synthesis, it depends on the response mode/syntesizer you use

There's a pretty good explanation here of response modes
https://docs.llamaindex.ai/en/stable/module_guides/deploying/query_engine/response_modes/?h=response+mode

@Logan M Yeah, that seems close to what I was looking for: https://docs.llamaindex.ai/en/stable/module_guides/deploying/query_engine/response_modes/?h=response+mode (contextual explanation w/out code). Thanks!