Do Retrieval-Augmented Language Models Adapt to Varying User Needs?

TL;DR

This paper proposes a new evaluation framework for Retrieval-Augmented Language Models that considers diverse user needs and retrieval scenarios, revealing how model performance varies with different user requirements and retrieval conditions.

Contribution

It introduces a systematic evaluation method for RALMs across multiple user need cases and context settings, highlighting the importance of user-centric assessments.

Findings

Restricting memory improves robustness in adversarial retrieval scenarios.

Model performance varies significantly with user needs and retrieval conditions.

Behavioral differences are mainly driven by the choice of model family.

Abstract

Recent advancements in Retrieval-Augmented Language Models (RALMs) have demonstrated their efficacy in knowledge-intensive tasks. However, existing evaluation benchmarks often assume a single optimal approach to leveraging retrieved information, failing to account for varying user needs. This paper introduces a novel evaluation framework that systematically assesses RALMs under three user need cases-Context-Exclusive, Context-First, and Memory-First-across three distinct context settings: Context Matching, Knowledge Conflict, and Information Irrelevant. By varying both user instructions and the nature of retrieved information, our approach captures the complexities of real-world applications where models must adapt to diverse user requirements. Through extensive experiments on multiple QA datasets, including HotpotQA, DisentQA, and our newly constructed synthetic URAQ dataset, we find that restricting memory usage improves robustness in adversarial retrieval conditions but decreases peak performance with ideal retrieval results and model family dominates behavioral differences. Our findings highlight the necessity of user-centric evaluations in the development of retrieval-augmented systems and provide insights into optimizing model performance across varied retrieval contexts. We will release our code and URAQ dataset upon acceptance of the paper.