Large Language Model Is Not a Good Few-shot Information Extractor, but a Good Reranker for Hard Samples!

TL;DR

This paper evaluates the effectiveness of Large Language Models (LLMs) in information extraction tasks, finding they are generally inferior to fine-tuned models but can enhance performance when used as rerankers for difficult samples.

Contribution

It introduces an adaptive filter-then-rerank paradigm combining SLMs and LLMs, demonstrating improved IE performance with minimal additional cost.

Findings

LLMs perform worse than fine-tuned models in most IE settings.

Using LLMs as rerankers improves extraction accuracy on hard samples.

The proposed method achieves an average 2.4% F1 gain across tasks.

Abstract

Large Language Models (LLMs) have made remarkable strides in various tasks. Whether LLMs are competitive few-shot solvers for information extraction (IE) tasks, however, remains an open problem. In this work, we aim to provide a thorough answer to this question. Through extensive experiments on nine datasets across four IE tasks, we demonstrate that current advanced LLMs consistently exhibit inferior performance, higher latency, and increased budget requirements compared to fine-tuned SLMs under most settings. Therefore, we conclude that LLMs are not effective few-shot information extractors in general. Nonetheless, we illustrate that with appropriate prompting strategies, LLMs can effectively complement SLMs and tackle challenging samples that SLMs struggle with. And moreover, we propose an adaptive filter-then-rerank paradigm to combine the strengths of LLMs and SLMs. In this paradigm, SLMs serve as filters and LLMs serve as rerankers. By prompting LLMs to rerank a small portion of difficult samples identified by SLMs, our preliminary system consistently achieves promising improvements (2.4% F1-gain on average) on various IE tasks, with an acceptable time and cost investment.