arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.03300cs.LGcs.AIcs.CLcs.CV

R1-SyntheticVL：生成模型的合成数据是否已为多模态大语言模型做好准备？

R1-SyntheticVL: Is Synthetic Data from Generative Models Ready for Multimodal Large Language Model?

Hong Kong Polytechnic University（香港理工大学）；Nanyang Technological University（南洋理工大学）；Tsinghua University（清华大学）；National University of Singapore（新加坡国立大学）

Jingyi Zhang, Tianyi Lin, Huanjin Yao, Xiang Lan, Shunyu Liu, Jiaxing Huang

备注：
ICML 2026 Camera Ready
阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出集体对抗数据合成（CADS）方法，通过集体智能和对抗学习自动生成高质量、多样且具有挑战性的多模态数据，用于增强多模态大语言模型（MLLM）在复杂现实任务中的性能。

摘要：

在这项工作中，我们旨在开发有效的数据合成技术，自主合成多模态训练数据，以增强MLLM解决复杂现实任务的能力。为此，我们提出了集体对抗数据合成（CADS），这是一种新颖且通用的方法，用于合成高质量、多样且具有挑战性的多模态数据。CADS的核心思想是利用集体智能确保高质量和多样化的生成，同时探索对抗学习以合成具有挑战性的样本，从而有效驱动模型改进。具体来说，CADS包含两个循环阶段：集体对抗数据生成（CAD-Generate）和集体对抗数据判断（CAD-Judge）。CAD-Generate利用集体知识共同生成新的多样化多模态数据，而CAD-Judge则协作评估合成数据的质量。此外，CADS引入了一种对抗上下文优化机制，以优化生成上下文，鼓励生成具有挑战性和高价值的数据。通过CADS，我们构建了MMSynthetic-20K并训练了我们的模型R1-SyntheticVL，该模型在多个基准测试中表现出优越的性能。

英文摘要：

In this work, we aim to develop effective data synthesis techniques that autonomously synthesize multimodal training data for enhancing MLLMs in solving complex real-world tasks. To this end, we propose Collective Adversarial Data Synthesis (CADS), a novel and general approach to synthesize high-quality, diverse and challenging multimodal data for MLLMs. The core idea of CADS is to leverage collective intelligence to ensure high-quality and diverse generation, while exploring adversarial learning to synthesize challenging samples for effectively driving model improvement. Specifically, CADS operates with two cyclic phases, i.e., Collective Adversarial Data Generation (CAD-Generate) and Collective Adversarial Data Judgment (CAD-Judge). CAD-Generate leverages collective knowledge to jointly generate new and diverse multimodal data, while CAD-Judge collaboratively assesses the quality of synthesized data. In addition, CADS introduces an Adversarial Context Optimization mechanism to optimize the generation context to encourage challenging and high-value data generation. With CADS, we construct MMSynthetic-20K and train our model R1-SyntheticVL, which demonstrates superior performance on various benchmarks.

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。

↑