Learning from Self Critique and Refinement for Faithful LLM Summarization

TL;DR

This paper introduces SCRPO, a self-supervised training method that enhances LLMs' faithfulness in summarization by leveraging their own critique and refinement capabilities, outperforming existing methods in faithfulness and efficiency.

Contribution

Proposes SCRPO, a novel self-supervised framework that uses LLMs' own critique to improve summarization faithfulness without extra test-time computation.

Findings

Outperforms state-of-the-art self-supervised methods in faithfulness metrics.

Achieves more faithful summaries compared to test-time refinement.

Maintains or improves overall summary quality.

Abstract

Large Language Models (LLMs) often suffer from hallucinations: output content that is not grounded in the input context, when performing long-form text generation tasks such as summarization. Prior works have shown that hallucinations can be reduced by iteratively critiquing and refining previously generated outputs using either the same model or a more powerful teacher model as the critique. However, these approaches either require additional test-time compute or assume access to more powerful teacher models, making them costly and less practical. In this work, we propose Self Critique and Refinement-based Preference Optimization (SCRPO), which is a self-supervised training framework that first constructs a preference dataset by leveraging the LLM's own critique and refinement capabilities, and then applies preference learning to improve the same LLM for faithful summarization. Experiments on three summarization benchmarks (XSUM CNNDM and SAMSum), demonstrate that our approach outperforms state-of-the-art self-supervised learning methods in terms of faithfulness metrics while either maintaining or improving other metrics that measure the overall quality of the summary. Moreover, compared to test-time refinement, our approach not only improves efficiency but also results in more faithful summaries.