KG_RAG
KG-RAG combines knowledge graphs with large language models to optimize prompt generation using a comprehensive biomedical knowledge base. The system facilitates efficient retrieval-augmented generation, crucial in tasks such as summarizing FDA drug information for conditions like Bardet-Biedl Syndrome. KG-RAG can be run through a straightforward setup, offering smooth interaction with models like GPT and Llama. It is benchmarked with the BiomixQA dataset to demonstrate its potential in advancing biomedical natural language processing.
