In this research, Retrieval-Augmented Large Language Models (LLMs) are explored as a way to improve response accuracy in tasks like Question-Answering (QA). Current approaches either add unnecessary computational complexity to simple queries or struggle with multi-step queries. To address this, a new adaptive QA framework is proposed, using a classifier to select the best strategy for LLMs based on query complexity. This model shows improved efficiency and accuracy in QA systems compared to other methods. The approach dynamically adjusts between different strategies, including iterative and single-step retrieval-augmented LLMs, based on the complexity of the query.
https://arxiv.org/abs/2403.14403