On the Role of Long-tail Knowledge in Retrieval Augmented Large Language Models

TL;DR

This paper emphasizes the importance of long-tail knowledge in retrieval augmented generation (RAG) for large language models, proposing a new detection method to improve efficiency and accuracy by focusing on less common, long-tail information.

Contribution

It introduces the GECE metric for detecting long-tail knowledge and demonstrates a targeted retrieval approach that enhances RAG performance and efficiency.

Findings

Over 4x speedup in inference time

Improved downstream task performance

Effective long-tail knowledge detection

Abstract

Retrieval augmented generation (RAG) exhibits outstanding performance in promoting the knowledge capabilities of large language models (LLMs) with retrieved documents related to user queries. However, RAG only focuses on improving the response quality of LLMs via enhancing queries indiscriminately with retrieved information, paying little attention to what type of knowledge LLMs really need to answer original queries more accurately. In this paper, we suggest that long-tail knowledge is crucial for RAG as LLMs have already remembered common world knowledge during large-scale pre-training. Based on our observation, we propose a simple but effective long-tail knowledge detection method for LLMs. Specifically, the novel Generative Expected Calibration Error (GECE) metric is derived to measure the ``long-tailness'' of knowledge based on both statistics and semantics. Hence, we retrieve relevant documents and infuse them into the model for patching knowledge loopholes only when the input query relates to long-tail knowledge. Experiments show that, compared to existing RAG pipelines, our method achieves over 4x speedup in average inference time and consistent performance improvement in downstream tasks.