Back to Search Start Over

ExaRanker: Explanation-Augmented Neural Ranker

Authors :
Ferraretto, Fernando
Laitz, Thiago
Lotufo, Roberto
Nogueira, Rodrigo
Publication Year :
2023

Abstract

Recent work has shown that inducing a large language model (LLM) to generate explanations prior to outputting an answer is an effective strategy to improve performance on a wide range of reasoning tasks. In this work, we show that neural rankers also benefit from explanations. We use LLMs such as GPT-3.5 to augment retrieval datasets with explanations and train a sequence-to-sequence ranking model to output a relevance label and an explanation for a given query-document pair. Our model, dubbed ExaRanker, finetuned on a few thousand examples with synthetic explanations performs on par with models finetuned on 3x more examples without explanations. Furthermore, the ExaRanker model incurs no additional computational cost during ranking and allows explanations to be requested on demand.

Details

Database :
arXiv
Publication Type :
Report
Accession number :
edsarx.2301.10521
Document Type :
Working Paper