I think training a model on your data will be in the end more effective ... take chatgpt for instance ... it's not an expert in python but it has "digested" enough documents about python to know his way around ... the "algorithm" that selects nodes is basically cosine similarity between your query and your nodes ... so yes, it sounds simplistic ... @Logan M what do you think ?
yea @stewie griffin fine-tuning may help in the end. But that's a looong road with a lot of complications.
In the meantime though, I still think llama-index can fufil your needs. We have things like the SubQuestionQueryEngine which can decompose complex queries into multiple questions, the RouterQueryEngine can route queries to the proper sub-index, etc.
