Sample generations from two models: RAG model using VedantaNY-10M and standard non-RAG LLM:
Sam Altman says that he believes in the absolute equivalence of Brahman and Atman. What does it mean? [RAG] [Non-RAG]
Does Swami speak about Wittgenstein's dissertation defense? [RAG] [Non-RAG]
Dreams feel absolutely real while dreaming. What is the guarantee that I am not fast asleep in my bed and dreaming right now? [RAG] [Non-RAG]
As our human evaluation shows, the RAG model outperforms the standard non-RAG model various axes, particularly in factuality, completeness and specificity and has a 82% preference rate. For more sample generations including those of the keyword-based hybrid RAG model, please refer to the paper.
Human evaluations are available here.