hi. is it possible to get 3 related source_texts when querying without using similarity_top_k?
If I include similarity_top_k, the response time is almost k times longer, so I would like to query the API only once.

Can I get LLM response times?

Can I merge multiple indices into one index?