LongDPO: Unlock Better Long-form Generation Abilities for LLMs via Critique-augmented Stepwise Information

ACL ARR 2025 February Submission1928 Authors

Abstract: Recent advancements in large language models (LLMs) have markedly improved their capacity to handle long text inputs; however, current models, including GPT-4o, still exhibit unsatisfactory performance in long-form generation. Generating high-quality long-form content still remains a significant challenge. In this paper, we present LongDPO, a novel approach designed to enhance long-form text generation through step-level supervision. By leveraging Monte Carlo Tree Search (MCTS) to collect stepwise preference pairs and employing a global memory pool to maintain factual accuracy, LongDPO effectively mitigates issues such as inconsistencies that are prevalent in long-context LLMs. Furthermore, we integrate critique-augmented generation to refine the selected preference pairs. Following the collection of stepwise preference pairs, we apply stepwise preference learning for fine-grained optimization. Experimental results demonstrate that our method enhances performance on long-form generation benchmarks (e.g. LongBench-Write) while maintaining nearly lossless performance on several general benchmarks.

Paper Type: Long

Research Area: Generation

Research Area Keywords: text-to-text generation, automatic evaluation, human evaluation

Contribution Types: Publicly available software and/or pre-trained models, Data resources

Languages Studied: English

Submission Number: 1928