反思性指令微调：缓解大型视觉-语言模型中的幻觉
Reflective Instruction Tuning: Mitigating Hallucinations in Large Vision-Language Models
AI总结：
针对大型视觉-语言模型幻觉问题，提出反思性指令微调方法，通过让模型预测响应正确与否的理由增强推理监督，并构建含115k条标注的REVERIE数据集，在多个基准上显著提升性能。
摘要：
大型视觉-语言模型（LVLMs）在各种视觉-语言任务中展现出令人期待的性能。然而，它们仍然容易产生幻觉，生成与视觉内容或指令不一致的输出。尽管已经提出了各种缓解策略，但它们往往忽略了一个导致幻觉的关键因素：训练过程中缺乏细粒度的推理监督。没有中间的推理步骤，模型可能会在指令和响应之间建立肤浅的捷径，无法内化固有的推理逻辑。为了解决这一挑战，我们提出了反思性指令微调，将理由学习整合到视觉指令微调中。与之前仅从响应中学习的方法不同，我们的方法要求模型预测理由，以证明响应为何正确或错误。这促进了模型对每个响应背后细粒度推理的更深层次参与，从而增强了模型的推理能力。为了支持这种方法，我们提出了REVERIE，这是第一个带有反思性理由标注的大规模指令微调数据集。REVERIE包含115k条机器生成的推理指令，每条指令都精心标注了对应的正确和混淆响应对，以及全面的理由，阐明每个响应正确或错误的依据。在多个LVLM基准上的实验结果表明，使用REVERIE数据集进行反思性指令微调相对于基线模型取得了显著的性能提升，证明了从理由中反思的有效性。项目页面位于https://zjr2000.github.io/projects/reverie。
英文摘要：
Large vision-language models (LVLMs) have shown promising performance on a variety of vision-language tasks. However, they remain susceptible to hallucinations, generating outputs misaligned with visual content or instructions. While various mitigation strategies have been proposed, they often neglect a key contributor to hallucinations: lack of fine-grained reasoning supervision during training. Without intermediate reasoning steps, models may establish superficial shortcuts between instructions and responses, failing to internalize the inherent reasoning logic. To address this challenge, we propose reflective instruction tuning, which integrates rationale learning into visual instruction tuning. Unlike previous methods that learning from responses only, our approach entails the model predicting rationales justifying why responses are correct or incorrect. This fosters a deeper engagement with the fine-grained reasoning underlying each response, thus enhancing the model's reasoning proficiency. To facilitate this approach, we propose REVERIE, the first large-scale instruction-tuning dataset with ReflEctiVE RatIonalE annotations. REVERIE comprises 115k machine-generated reasoning instructions, each meticulously annotated with a corresponding pair of correct and confusing responses, alongside comprehensive rationales elucidating the justification behind the correctness or erroneousness of each response. Experimental results on multiple LVLM benchmarks reveal that reflective instruction tuning with the REVERIE dataset yields noticeable performance gain over the baseline model, demonstrating the effectiveness of reflecting from the rationales. Project page is at https://zjr2000.github.io/projects/reverie.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
