RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Yu, Yue, Ping, Wei, Liu, Zihan, Wang, Boxin, You, Jiaxuan, Zhang, Chao, Shoeybi, Mohammad, Catanzaro, Bryan
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910510192001024
author Yu, Yue
Ping, Wei
Liu, Zihan
Wang, Boxin
You, Jiaxuan
Zhang, Chao
Shoeybi, Mohammad
Catanzaro, Bryan
author_facet Yu, Yue
Ping, Wei
Liu, Zihan
Wang, Boxin
You, Jiaxuan
Zhang, Chao
Shoeybi, Mohammad
Catanzaro, Bryan
contents Large language models (LLMs) typically utilize the top-k contexts from a retriever in retrieval-augmented generation (RAG). In this work, we propose a novel instruction fine-tuning framework RankRAG, which instruction-tunes a single LLM for the dual purpose of context ranking and answer generation in RAG. In particular, the instruction-tuned LLMs work surprisingly well by adding a small fraction of ranking data into the training blend, and outperform existing expert ranking models, including the same LLM exclusively fine-tuned on a large amount of ranking data. For generation, we compare our model with many strong baselines, including GPT-4-0613, GPT-4-turbo-2024-0409, and ChatQA-1.5, an open-sourced model with the state-of-the-art performance on RAG benchmarks. Specifically, our Llama3-RankRAG significantly outperforms Llama3-ChatQA-1.5 and GPT-4 models on nine knowledge-intensive benchmarks. In addition, it also performs comparably to GPT-4 on five RAG benchmarks in the biomedical domain without instruction fine-tuning on biomedical data, demonstrating its superb capability for generalization to new domains.
format Preprint
id arxiv_https___arxiv_org_abs_2407_02485
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs
Yu, Yue
Ping, Wei
Liu, Zihan
Wang, Boxin
You, Jiaxuan
Zhang, Chao
Shoeybi, Mohammad
Catanzaro, Bryan
Computation and Language
Artificial Intelligence
Information Retrieval
Machine Learning
Large language models (LLMs) typically utilize the top-k contexts from a retriever in retrieval-augmented generation (RAG). In this work, we propose a novel instruction fine-tuning framework RankRAG, which instruction-tunes a single LLM for the dual purpose of context ranking and answer generation in RAG. In particular, the instruction-tuned LLMs work surprisingly well by adding a small fraction of ranking data into the training blend, and outperform existing expert ranking models, including the same LLM exclusively fine-tuned on a large amount of ranking data. For generation, we compare our model with many strong baselines, including GPT-4-0613, GPT-4-turbo-2024-0409, and ChatQA-1.5, an open-sourced model with the state-of-the-art performance on RAG benchmarks. Specifically, our Llama3-RankRAG significantly outperforms Llama3-ChatQA-1.5 and GPT-4 models on nine knowledge-intensive benchmarks. In addition, it also performs comparably to GPT-4 on five RAG benchmarks in the biomedical domain without instruction fine-tuning on biomedical data, demonstrating its superb capability for generalization to new domains.
title RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs
topic Computation and Language
Artificial Intelligence
Information Retrieval
Machine Learning
url https://arxiv.org/abs/2407.02485