ImageNet-Think-250K：面向视觉语言模型多模态推理的大规模合成数据集
ImageNet-Think-250K: A Large-Scale Synthetic Dataset for Multimodal Reasoning for Vision Language Models
AI总结：
本研究推出基于ImageNet21k的25万张图像构建的ImageNet-Think多模态推理数据集，由两款前沿VLM生成每组图像的两组思维-答案序列对，记录逐步推理过程，将公开以支撑多模态推理VLM的研发与机制研究。
摘要：
我们构建了ImageNet-Think多模态推理数据集，旨在助力开发具备显式推理能力的视觉语言模型（VLM）。该数据集基于ImageNet21k数据集中的25万张图像构建，提供结构化的思维令牌及对应答案。我们的合成数据集由两款前沿VLM生成：GLM-4.1V-9B-Thinking和Kimi-VL-A3B-Thinking-2506。每张图像配有两组思维-答案序列对，可为多模态推理模型的训练与评估提供资源。我们记录了VLM的逐步推理过程及最终描述性答案。构建该数据集的目标是推动开发更鲁棒的VLM，同时助力更深入地理解多模态推理机制。数据集及评估基准将公开发布，以支持推理/思维型多模态VLM领域的研究。
英文摘要：
We develop ImageNet-Think, a multimodal reasoning dataset designed to aid the development of Vision Language Models (VLMs) with explicit reasoning capabilities. Our dataset is built on 250,000 images from ImageNet21k dataset, providing structured thinking tokens and corresponding answers. Our synthetic dataset is generated by two state-of-the-art VLMs: GLM-4.1V-9B-Thinking and Kimi-VL-A3B-Thinking-2506. Each image is accompanied by two pairs of thinking-answer sequences, creating a resource for training and evaluating multimodal reasoning models. We capture the step-by-step reasoning process of VLMs and the final descriptive answers. Our goal with this dataset is to enable the development of more robust VLMs while contributing to the broader understanding of multimodal reasoning mechanisms. The dataset and evaluation benchmarks will be publicly available to aid research in reasoning/thinking multimodal VLMs.
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。
