LLMs have transformed how humans search for information for everyday tasks.
Although well suited for general scenarios, LLMs hallucinate and produce irrelevant information when queried on specialized knowledge such as medicine, law, finance, and more.
They also do not give up-to-date information in these constantly updating fields and offer simplistic responses without considering novel insights or discoveries.
LLMs can also not access out-of-the-box private and specialized data related to a field unless they are fine-tuned in it. However, fine-tuning is a complex process involving domain expertise, considerable time, and computational resources.
To combat this issue, Retrieval Augmented Generation (RAG) was introduced in 2021. This method lets LLMs answer user queries using specialized private datasets without requiring any fine-tuning.
The process was made even more accurate in early 2024 using Graph Retrieval-Augmented Generation (GRAG).
Finally, we have MedG…
