go back

Accessing Knowledge using Retrieval Augmented Generation

Felix Ocker and Julian Eggert, "Accessing Knowledge using Retrieval Augmented Generation", Honda Technical Forum 2023, 2023.

Abstract

Language Models (LMs) provide an intuitive interface for humans via Natural Language. However, they hallucinate very convincingly and do not have access to proprietary data. This presentation gives insights into Retrieval Augmented Generation (RAG) as a technology for realizing the LM experience for large amounts of proprietary data. We present the underlying architecture, results achieved with an HRI-EU internal prototype for the TikiWiki, the "wiki bot", and give an outlook regarding the technology's impact and future research directions.



Download Bibtex file Per Mail Request

Search