vix.ing · top · new · best · stats

Large Language Models Struggle to Learn Long-Tail Knowledge

2022/11/15 by Nikhil Kandpal, Kandpal, Nikhil, Haikang Deng +7 · 2 voices · 122 citations
Computer Science · #Artificial intelligence #Code (set theory) #Computer science #Expert finding and Q&A systems #Information retrieval #Language model #Natural Language Processing Techniques #Natural language processing #Question answering #Text corpus #The Internet #Topic Modeling #Training set #World Wide Web

paper · pdf · doi:10.48550/arxiv.2211.08411

published in arXiv (Cornell University) (Cornell University)

openalex publication_date 2022/11/15 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

The Internet contains a wealth of knowledge -- from the birthdays of historical figures to tutorials on how to code -- all of which may be learned by language models. However, while certain pieces of information are ubiquitous on the web, others appear extremely rarely. In this paper, we study the relationship between the knowledge memorized by large language models and the information in pre-training datasets scraped from the web. In particular, we show that a language model's ability to answer a fact-based question relates to how many documents associated with that question were seen during pre-training. We identify these relevant documents by entity linking pre-training datasets and counting documents that contain the same entities as a given question-answer pair. Our results demonstrate strong correlational and causal relationships between accuracy and relevant document count for numerous question answering datasets (e.g., TriviaQA), pre-training corpora (e.g., ROOTS), and model sizes (e.g., 176B parameters). Moreover, while larger models are better at learning long-tail knowledge, we estimate that today's models must be scaled by many orders of magnitude to reach competitive QA performance on questions with little support in the pre-training data. Finally, we show that retrieval-augmentation can reduce the dependence on relevant pre-training information, presenting a promising approach for capturing the long-tail.

Cited by

Discussions

Related