Fine-tuning Large Language Models for Improving Factuality in Legal Question Answering

TL;DR

This paper introduces a new benchmark and metrics for evaluating hallucinations in legal question answering by large language models, and proposes a novel mitigation method that significantly reduces hallucination rates.

Contribution

It presents a new benchmark, metrics, and a hallucination mitigation technique combining behavior cloning and HIPO for legal LLM QA.

Findings

Significant improvements in hallucination-related metrics.

Effective reduction in hallucination rates in legal QA.

Enhanced answer factuality and relevance.

Abstract

Hallucination, or the generation of incorrect or fabricated information, remains a critical challenge in large language models (LLMs), particularly in high-stake domains such as legal question answering (QA). In order to mitigate the hallucination rate in legal QA, we first introduce a benchmark called LegalHalBench and three automatic metrics to evaluate the common hallucinations when LLMs answer legal questions. We then propose a hallucination mitigation method that integrates behavior cloning and a novel Hard Sample-aware Iterative Direct Preference Optimization (HIPO). We conduct extensive real-data experiments to validate the effectiveness of our approach. Our results demonstrate remarkable improvements in various metrics, including the newly proposed Non-Hallucinated Statute Rate, Statute Relevance Rate, Legal Claim Truthfulness, as well as traditional metrics such as METEOR, BERTScore, ROUGE-L, and win rates.