lmms-lab/LLaVA-NeXT-Interleave-Bench

LLaVA-Interleave Bench是一个全面的多图像数据集，收集自公共数据集或通过GPT-4V API生成，旨在评估大型多模态模型（LMMs）的交错多图像推理能力。数据集于2024年4月收集，并于2024年6月发布。它包含多个配置，每个配置有不同的特征和测试集，主要用于计算机视觉、自然语言处理、机器学习和人工智能领域的研究人员和爱好者。

LLaVA-Interleave Bench is a comprehensive set of multi-image datasets collected from public datasets or generated by the GPT-4V API. It is designed to evaluate the interleaved multi-image reasoning capabilities of large multimodal models (LMMs). The dataset includes multiple configurations such as in_domain, multi_view_in_domain, and out_of_domain, each with specific features and splits. The dataset contains images, questions, and answers, and is primarily intended for research in areas like computer vision, natural language processing, and machine learning. The dataset was collected in April 2024 and released in June 2024 under the Creative Commons Attribution 4.0 International license.