Reimagining Retrieval Augmented Language Models for Answering Queries (2306.01061v1)

Abstract: We present a reality check on LLMs and inspect the promise of retrieval augmented LLMs in comparison. Such LLMs are semi-parametric, where models integrate model parameters and knowledge from external data sources to make their predictions, as opposed to the parametric nature of vanilla LLMs. We give initial experimental findings that semi-parametric architectures can be enhanced with views, a query analyzer/planner, and provenance to make a significantly more powerful system for question answering in terms of accuracy and efficiency, and potentially for other NLP tasks