With the rapid advancement of multimodal large language models (MLLMs), their evaluation has become increasingly comprehensive. However, understanding long multimodal content, as a foundational ability for real-world applications, remains underexplored. In this work, we present Needle In A Multimodal Haystack (MM-NIAH), the first benchmark specifically designed to systematically evaluate the capability of existing MLLMs to comprehend long multimodal documents. Our benchmark includes three types of evaluation tasks: multimodal retrieval, counting, and reasoning. In each task, the model is required to answer the questions according to different key information scattered throughout the given multimodal document. Evaluating the leading MLLMs on MM-NIAH, we observe that existing models still have significant room for improvement on these tasks, especially on vision-centric evaluation. We hope this work can provide a platform for further research on long multimodal document comprehension and contribute to the advancement of MLLMs. Code and benchmark are released at https://github.com/OpenGVLab/MM-NIAH.
@article{arxiv.2406.07230,
title = {Needle In A Multimodal Haystack},
author = {Weiyun Wang and Shuibo Zhang and Yiming Ren and Yuchen Duan and Tiantong Li and Shuo Liu and Mengkang Hu and Zhe Chen and Kaipeng Zhang and Lewei Lu and Xizhou Zhu and Ping Luo and Yu Qiao and Jifeng Dai and Wenqi Shao and Wenhai Wang},
journal= {arXiv preprint arXiv:2406.07230},
year = {2024}
}
Comments
Accepted to NeurIPS 2024 Track Datasets and Benchmarks