ReFusion: Improving Natural Language Understanding with Computation-Efficient Retrieval Representation Fusion

TL;DR

ReFusion introduces a computation-efficient method for integrating retrieval representations into language models, enhancing performance in knowledge-intensive and non-knowledge-intensive tasks without increasing input length.

Contribution

ReFusion presents a novel bi-level optimization approach that fuses retrieval representations directly into model hidden states, reducing computational costs compared to concatenation methods.

Findings

Achieves superior performance in various NKI tasks.

Reduces computational demands of retrieval augmentation.

Demonstrates robustness across multiple tasks.

Abstract

Retrieval-based augmentations (RA) incorporating knowledge from an external database into language models have greatly succeeded in various knowledge-intensive (KI) tasks. However, integrating retrievals in non-knowledge-intensive (NKI) tasks is still challenging. Existing works focus on concatenating retrievals with inputs to improve model performance. Unfortunately, the use of retrieval concatenation-based augmentations causes an increase in the input length, substantially raising the computational demands of attention mechanisms. This paper proposes a new paradigm of RA named \textbf{ReFusion}, a computation-efficient Retrieval representation Fusion with bi-level optimization. Unlike previous works, ReFusion directly fuses the retrieval representations into the hidden states of models. Specifically, ReFusion leverages an adaptive retrieval integrator to seek the optimal combination of the proposed ranking schemes across different model layers. Experimental results demonstrate that the proposed ReFusion can achieve superior and robust performance in various NKI tasks.