KnowTuning: Knowledge-aware Fine-tuning for Large Language Models

TL;DR

KnowTuning enhances large language models' ability to leverage knowledge effectively by training them to identify and distinguish reliable knowledge, improving factual accuracy and logical consistency in knowledge-intensive NLP tasks.

Contribution

The paper introduces a novel knowledge-aware fine-tuning method that improves LLMs' knowledge awareness through fine-grained augmentation and coarse-grained comparison stages.

Findings

Improves factual accuracy in LLMs on QA tasks.

Reduces factual error rate in generated responses.

Effective across various LLM sizes and datasets.

Abstract

Despite their success at many natural language processing (NLP) tasks, large language models still struggle to effectively leverage knowledge for knowledge-intensive tasks, manifesting limitations such as generating incomplete, non-factual, or illogical answers. These limitations stem from inadequate knowledge awareness of LLMs during vanilla fine-tuning. To address these problems, we propose a knowledge-aware fine-tuning (KnowTuning) method to improve fine-grained and coarse-grained knowledge awareness of LLMs. We devise a fine-grained knowledge augmentation stage to train LLMs to identify difficult fine-grained knowledge in answers. We also propose a coarse-grained knowledge comparison stage to train LLMs to distinguish between reliable and unreliable knowledge, in three aspects: completeness, factuality, and logicality. Extensive experiments on both generic and medical question answering (QA) datasets confirm the effectiveness of KnowTuning, through automatic and human evaluations, across various sizes of LLMs. We further verify that KnowTuning generates more facts with less factual error rate under fine-grained facts evaluation.