Saved in:
Bibliographic Details
Main Authors: Yang, Zhuoyi, Song, Yurun, Ahmed, Iftekhar, Harris, Ian
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.07054
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911368458797056
author Yang, Zhuoyi
Song, Yurun
Ahmed, Iftekhar
Harris, Ian
author_facet Yang, Zhuoyi
Song, Yurun
Ahmed, Iftekhar
Harris, Ian
contents Multi-hop question answering is widely used to evaluate the reasoning capabilities of large language models (LLMs), as it requires integrating multiple pieces of supporting knowledge to arrive at a correct answer. While prior work has explored different mechanisms for providing knowledge to LLMs, such as finetuning and retrieval-augmented generation (RAG), their relative effectiveness for multi-hop question answering remains insufficiently understood, particularly when the required knowledge is temporally novel. In this paper, we systematically compare parametric and non-parametric knowledge injection methods for open-domain multi-hop question answering. We evaluate unsupervised fine-tuning (continual pretraining), supervised fine-tuning, and retrieval-augmented generation across three 7B-parameter open-source LLMs. Experiments are conducted on two benchmarks: QASC, a standard multi-hop science question answering dataset, and a newly constructed dataset of over 10,000 multi-hop questions derived from Wikipedia events in 2024, designed to test knowledge beyond the models' pretraining cutoff. Our results show that unsupervised fine-tuning provides only limited gains over base models, suggesting that continual pretraining alone is insufficient for improving multi-hop reasoning accuracy. In contrast, retrieval-augmented generation yields substantial and consistent improvements, particularly when answering questions that rely on temporally novel information. Supervised fine-tuning achieves the highest overall accuracy across models and datasets. These findings highlight fundamental differences in how knowledge injection mechanisms support multi-hop question answering and underscore the importance of retrieval-based methods when external or compositional knowledge is required.
format Preprint
id arxiv_https___arxiv_org_abs_2601_07054
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Fine-Tuning vs. RAG for Multi-Hop Question Answering with Novel Knowledge
Yang, Zhuoyi
Song, Yurun
Ahmed, Iftekhar
Harris, Ian
Computation and Language
Machine Learning
Multi-hop question answering is widely used to evaluate the reasoning capabilities of large language models (LLMs), as it requires integrating multiple pieces of supporting knowledge to arrive at a correct answer. While prior work has explored different mechanisms for providing knowledge to LLMs, such as finetuning and retrieval-augmented generation (RAG), their relative effectiveness for multi-hop question answering remains insufficiently understood, particularly when the required knowledge is temporally novel. In this paper, we systematically compare parametric and non-parametric knowledge injection methods for open-domain multi-hop question answering. We evaluate unsupervised fine-tuning (continual pretraining), supervised fine-tuning, and retrieval-augmented generation across three 7B-parameter open-source LLMs. Experiments are conducted on two benchmarks: QASC, a standard multi-hop science question answering dataset, and a newly constructed dataset of over 10,000 multi-hop questions derived from Wikipedia events in 2024, designed to test knowledge beyond the models' pretraining cutoff. Our results show that unsupervised fine-tuning provides only limited gains over base models, suggesting that continual pretraining alone is insufficient for improving multi-hop reasoning accuracy. In contrast, retrieval-augmented generation yields substantial and consistent improvements, particularly when answering questions that rely on temporally novel information. Supervised fine-tuning achieves the highest overall accuracy across models and datasets. These findings highlight fundamental differences in how knowledge injection mechanisms support multi-hop question answering and underscore the importance of retrieval-based methods when external or compositional knowledge is required.
title Fine-Tuning vs. RAG for Multi-Hop Question Answering with Novel Knowledge
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2601.07054