Evaluating the External and Parametric Knowledge Fusion of Large Language Models

TL;DR

This paper systematically investigates how large language models fuse external and intrinsic parametric knowledge, revealing that strengthening parametric knowledge improves integration but faces challenges in memorization and boundary detection.

Contribution

It introduces a novel framework for analyzing knowledge fusion scenarios in LLMs and provides a systematic pipeline for controlled experiments on knowledge integration.

Findings

Enhancing parametric knowledge improves fusion capabilities.

Persistent challenges in memorizing and eliciting parametric knowledge.

Identified boundaries of parametric knowledge in LLMs.

Abstract

Integrating external knowledge into large language models (LLMs) presents a promising solution to overcome the limitations imposed by their antiquated and static parametric memory. Prior studies, however, have tended to over-reliance on external knowledge, underestimating the valuable contributions of an LLMs' intrinsic parametric knowledge. The efficacy of LLMs in blending external and parametric knowledge remains largely unexplored, especially in cases where external knowledge is incomplete and necessitates supplementation by their parametric knowledge. We propose to deconstruct knowledge fusion into four distinct scenarios, offering the first thorough investigation of LLM behavior across each. We develop a systematic pipeline for data construction and knowledge infusion to simulate these fusion scenarios, facilitating a series of controlled experiments. Our investigation reveals that enhancing parametric knowledge within LLMs can significantly bolster their capability for knowledge integration. Nonetheless, we identify persistent challenges in memorizing and eliciting parametric knowledge, and determining parametric knowledge boundaries. Our findings aim to steer future explorations on harmonizing external and parametric knowledge within LLMs.