You're currently viewing an old version of this dataset. To see the current version, click here.

Multimodal Robustness Benchmark

The MMR benchmark is designed to evaluate MLLMs' comprehension of visual content and robustness against misleading questions, ensuring models truly leverage multimodal inputs rather than relying solely on textual reasoning. The MMR-data is generated to enhance MLLMs' understanding capability and robustness by providing a training set with paired positive and negative visual question-answer samples.

Data and Resources

Cite this as

Yexin Liu, Zhengyang Liang, Yueze Wang, Muyang He, Jian Li, Bo Zhao (2024). Dataset: Multimodal Robustness Benchmark. https://doi.org/10.57702/hvw2a0mw

DOI retrieved: December 16, 2024

Additional Info

Field Value
Created December 16, 2024
Last update December 16, 2024
Defined In https://doi.org/10.48550/arXiv.2406.10638
Author Yexin Liu
More Authors
Zhengyang Liang
Yueze Wang
Muyang He
Jian Li
Bo Zhao
Homepage https://github.com/BAAI-DCAI/Multimodal-Robustness-Benchmark