I'm using this link to a preprint paper (16 page PDF) to introduce and define the concept of Retrieval-Augmented Generation (RAG). "Conceptually, RAG in educational contexts supports the accuracy of LLM-based QA by utilizing external knowledge sources such as syllabi, workbooks, and handouts. RAG injects information from a retrieval knowledge base into the LLM during inference (e.g., via prompts) to enhance the generated content through knowledge integration." The proposal in this specific paper is that structured information, such as a knowledge graph (KG), might serve the same purpose, only better. "KG-based RAG approach can enhance the groundedness and faithfulness of both the retrieved documents and the generated context."
Today: 26 Total: 576 [Direct link] [Share]
