Enhancing RAG with Active Learning on Conversation Records: Reject Incapables and Answer Capables

TL;DR

This paper enhances retrieval-augmented generation by applying active learning to conversation records, effectively selecting samples for annotation to reduce hallucinations in large language models.

Contribution

It introduces AL4RAG, a novel active learning framework tailored for RAG, including a new sample distance metric, to improve dataset quality and model performance.

Findings

AL4RAG outperforms baseline methods in multiple metrics.

The new distance metric improves sample selection for RAG.

Active learning reduces annotation costs while maintaining high performance.

Abstract

Retrieval-augmented generation (RAG) is a key technique for leveraging external knowledge and reducing hallucinations in large language models (LLMs). However, RAG still struggles to fully prevent hallucinated responses. To address this, it is essential to identify samples prone to hallucination or guide LLMs toward correct responses, which experts then annotate to develop high-quality datasets for refining LLMs. However, the growing scarcity of such datasets makes their creation challenging. This paper proposes using the vast amount of conversations from widespread LLM usage to build these datasets, training LLMs to avoid hallucination-prone questions while accurately responding to manageable ones. Given the impracticality of expert-annotating all conversation records, the paper introduces AL4RAG, which uses active learning to select the most suitable conversation samples for annotation, optimizing performance within an annotation budget. Additionally, recognizing that traditional active learning methods are not fully compatible with RAG due to unsuitable distance metrics, we develop a novel sample distance measurement for RAG active learning. Extensive experiments show that our method consistently outperforms baselines across multiple metrics.