When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories

TL;DR

This paper investigates the limitations of large language models in memorizing factual knowledge, especially less popular facts, and proposes a retrieval-augmented approach that improves performance and efficiency.

Contribution

It provides a comprehensive analysis of LMs' memorization capabilities and introduces a simple retrieval method that enhances factual recall and reduces inference costs.

Findings

Retrieval-augmented LMs outperform larger LMs in factual knowledge tasks.

Scaling does not significantly improve memorization of long-tail facts.

Retrieval-augmented approach reduces inference costs while maintaining high performance.

Abstract

Despite their impressive performance on diverse tasks, large language models (LMs) still struggle with tasks requiring rich world knowledge, implying the limitations of relying solely on their parameters to encode a wealth of world knowledge. This paper aims to understand LMs' strengths and limitations in memorizing factual knowledge, by conducting large-scale knowledge probing experiments of 10 models and 4 augmentation methods on PopQA, our new open-domain QA dataset with 14k questions. We find that LMs struggle with less popular factual knowledge, and that scaling fails to appreciably improve memorization of factual knowledge in the long tail. We then show that retrieval-augmented LMs largely outperform orders of magnitude larger LMs, while unassisted LMs remain competitive in questions about high-popularity entities. Based on those findings, we devise a simple, yet effective, method for powerful and efficient retrieval-augmented LMs, which retrieves non-parametric memories only when necessary. Experimental results show that this significantly improves models' performance while reducing the inference costs.