Evaluating the Retrieval Robustness of Large Language Models

TL;DR

This paper assesses how robust large language models are when using retrieval-augmented generation, revealing high robustness but also limitations in fully leveraging retrieved information across various models and strategies.

Contribution

It introduces a benchmark and metrics to evaluate retrieval robustness in LLMs, providing comprehensive insights into their performance and limitations.

Findings

LLMs show surprisingly high retrieval robustness

Imperfect robustness limits full benefit of RAG

Different models and strategies exhibit varying robustness levels

Abstract

Retrieval-augmented generation (RAG) generally enhances large language models' (LLMs) ability to solve knowledge-intensive tasks. But RAG may also lead to performance degradation due to imperfect retrieval and the model's limited ability to leverage retrieved content. In this work, we evaluate the robustness of LLMs in practical RAG setups (henceforth retrieval robustness). We focus on three research questions: (1) whether RAG is always better than non-RAG; (2) whether more retrieved documents always lead to better performance; (3) and whether document orders impact results. To facilitate this study, we establish a benchmark of 1500 open-domain questions, each with retrieved documents from Wikipedia. We introduce three robustness metrics, each corresponds to one research question. Our comprehensive experiments, involving 11 LLMs and 3 prompting strategies, reveal that all of these LLMs exhibit surprisingly high retrieval robustness; nonetheless, different degrees of imperfect robustness hinders them from fully utilizing the benefits of RAG.