Enhancing Document-Level Question Answering via Multi-Hop Retrieval-Augmented Generation with LLaMA 3

TL;DR

This paper introduces a retrieval-augmented generation framework based on LLaMA 3 that enhances complex document-level question answering by improving multi-hop reasoning and contextual understanding, leading to more accurate and coherent responses.

Contribution

The paper proposes a novel RAG framework with multi-hop reasoning and context fusion built on LLaMA 3, advancing the capabilities of retrieval-augmented question answering systems.

Findings

Outperforms existing retrieval-augmented baselines

Improves accuracy in multi-hop reasoning tasks

Enhances coherence in generated answers

Abstract

This paper presents a novel Retrieval-Augmented Generation (RAG) framework tailored for complex question answering tasks, addressing challenges in multi-hop reasoning and contextual understanding across lengthy documents. Built upon LLaMA 3, the framework integrates a dense retrieval module with advanced context fusion and multi-hop reasoning mechanisms, enabling more accurate and coherent response generation. A joint optimization strategy combining retrieval likelihood and generation cross-entropy improves the model's robustness and adaptability. Experimental results show that the proposed system outperforms existing retrieval-augmented and generative baselines, confirming its effectiveness in delivering precise, contextually grounded answers.