A + B: A General Generator-Reader Framework for Optimizing LLMs to Unleash Synergy Potential

TL;DR

This paper introduces a versatile 'A + B' framework that combines different foundation models to optimize large language models, enhancing their ability to utilize source knowledge and improve performance in complex scenarios.

Contribution

It formalizes a general framework for combining models, explores different model functionalities, and extends the approach to incorporate external knowledge via continuous learning.

Findings

Model combinations outperform single models in complex tasks

Different LLM versions are suitable for generator and reader roles

Framework effectively integrates external knowledge into LLMs

Abstract

Retrieval-Augmented Generation (RAG) is an effective solution to supplement necessary knowledge to large language models (LLMs). Targeting its bottleneck of retriever performance, "generate-then-read" pipeline is proposed to replace the retrieval stage with generation from the LLM itself. Although promising, this research direction is underexplored and still cannot work in the scenario when source knowledge is given. In this paper, we formalize a general "A + B" framework with varying combinations of foundation models and types for systematic investigation. We explore the efficacy of the base and chat versions of LLMs and found their different functionalities suitable for generator A and reader B, respectively. Their combinations consistently outperform single models, especially in complex scenarios. Furthermore, we extend the application of the "A + B" framework to scenarios involving source documents through continuous learning, enabling the direct integration of external knowledge into LLMs. This approach not only facilitates effective acquisition of new knowledge but also addresses the challenges of safety and helpfulness post-adaptation. The paper underscores the versatility of the "A + B" framework, demonstrating its potential to enhance the practical application of LLMs across various domains.