Advanced Optimization for RAG

[hiriki]

Have you considered some advanced optimizations for RAG, such as Hyde, decomposition COT, etc.? In the relevant code, I only saw optimizations related to basic concepts, but not those dealing with some more difficult structures, such as QA pairs in COT or the concept of Hyde adding documents before retrieval. Is it because you think adding these would make the entire project more chaotic and burdensome that these are left to be implemented by the developers themselves?