arXiv 2604.102282026-05-28cs.AI

SVSR：一种用于多模态推理的自我验证与自我修正范式

SVSR: A Self-Verification and Self-Rectification Paradigm for Multimodal Reasoning

South China Agricultural University（华南农业大学）；University of Glasgow（格拉斯哥大学）；University of Electronic Science and Technology of China（电子科技大学）；Monash University（莫纳什大学）；University of Science and Technology of China（中国科学技术大学）；National University of Defense Technology（国防科技大学）；Renmin University of China（中国人民大学）；South China Normal University（华南师范大学）

Zhe Qian, Nianbing Su, Zhonghua Wang, Hebei Li, Zhongxing Xu, Yueying Li, Fei Luo, Zhuohan Ouyang, Yanbiao Ma

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出SVSR框架，通过三阶段训练（统一偏好数据集构建、冷启动监督微调、半在线直接偏好优化）将自我验证与自我修正显式集成到多模态推理流程中，提升复杂视觉理解和多模态推理的鲁棒性与可靠性。

摘要：

当前多模态模型常存在浅层推理问题，导致因不完整或不一致的思维过程而产生错误。为解决这一局限，我们提出自我验证与自我修正（SVSR）统一框架，将自我验证和自我修正显式集成到模型的推理流程中，显著提升复杂视觉理解和多模态推理任务的鲁棒性与可靠性。SVSR基于一种新颖的三阶段训练范式。首先，通过精炼预训练视觉语言模型的推理轨迹，结合前向和后向推理嵌入自我反思信号，构建高质量统一偏好数据集。其次，在该数据集上进行冷启动监督微调，学习结构化、多步推理行为。第三，应用半在线直接偏好优化（Semi-online DPO）过程，通过强大的教师VLM筛选的高质量模型生成推理轨迹持续增强训练语料。该流程使模型能够学习、激发并精炼其自我验证与自我修正能力。跨多个基准的广泛实验表明，SVSR提升了推理准确性，并增强了对未见任务和问题类型的泛化能力。值得注意的是，经过显式自我反思推理训练后，模型还展现出改进的隐式推理能力，即使在没有显式推理轨迹的情况下也优于强基线。这些结果凸显了SVSR在构建更可靠、内省且认知对齐的多模态系统方面的潜力。

英文摘要：

Current multimodal models often suffer from shallow reasoning, leading to errors caused by incomplete or inconsistent thought processes. To address this limitation, we propose Self-Verification and Self-Rectification (SVSR), a unified framework that explicitly integrates self-verification and self-rectification into the model's reasoning pipeline, substantially improving robustness and reliability in complex visual understanding and multimodal reasoning tasks. SVSR is built on a novel three-stage training paradigm. First, we construct a high-quality unified preference dataset by refining reasoning traces from pre-trained vision-language models, incorporating both forward and backward reasoning to embed self-reflective signals. Second, we perform cold-start supervised fine-tuning on this dataset to learn structured, multi-step reasoning behaviors. Third, we apply a Semi-online Direct Preference Optimization (Semi-online DPO) process, continuously augmenting the training corpus with high-quality, model-generated reasoning traces filtered by a powerful teacher VLM. This pipeline enables the model to learn, elicit, and refine its ability to self-verify and self-rectify. Extensive experiments across diverse benchmarks demonstrate that SVSR improves reasoning accuracy and enables stronger generalization to unseen tasks and question types. Notably, once trained with explicit self-reflective reasoning, the model also exhibits improved implicit reasoning ability, outperforming strong baselines even when no explicit reasoning traces are provided. These results highlight the potential of SVSR for building more dependable, introspective, and cognitively aligned multimodal systems.

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。