InFi-Check: Interpretable and Fine-Grained Fact-Checking of LLMs

TL;DR

InFi-Check is a novel framework that enables interpretable, fine-grained fact-checking of LLM outputs by generating explicit evidence, error classifications, justifications, and corrections, improving trustworthiness.

Contribution

The paper introduces a controlled data synthesis pipeline, a large-scale training dataset, and a new model, InFi-Checker, for interpretable, fine-grained fact-checking of LLMs.

Findings

InFi-Checker achieves state-of-the-art results on InFi-Check-FG.

It generalizes well across various downstream tasks.

It significantly enhances the utility and trustworthiness of factuality evaluation.

Abstract

Large language models (LLMs) often hallucinate, yet most existing fact-checking methods treat factuality evaluation as a binary classification problem, offering limited interpretability and failing to capture fine-grained error types. In this paper, we introduce InFi-Check, a framework for interpretable and fine-grained fact-checking of LLM outputs. Specifically, we first propose a controlled data synthesis pipeline that generates high-quality data featuring explicit evidence, fine-grained error type labels, justifications, and corrections. Based on this, we further construct large-scale training data and a manually verified benchmark InFi-Check-FG for fine-grained fact-checking of LLM outputs. Building on these high-quality training data, we further propose InFi-Checker, which can jointly provide supporting evidence, classify fine-grained error types, and produce justifications along with corrections. Experiments show that InFi-Checker achieves state-of-the-art performance on InFi-Check-FG and strong generalization across various downstream tasks, significantly improving the utility and trustworthiness of factuality evaluation.