Devil in the Number: Towards Robust Multi-modality Data Filter

The dataset used in the paper is a web-scale dataset for training a vision-language model. The dataset contains text-image pairs, and the authors propose a novel filter to remove redundant information such as numbers and bracketed content.

Data and Resources

Cite this as

Yichen Xu, Zihan Xu, Wenhao Chai, Zhonghan Zhao, Enxin Song, Gaoang Wang (2024). Dataset: Devil in the Number: Towards Robust Multi-modality Data Filter. https://doi.org/10.57702/wrxz766y

DOI retrieved: December 2, 2024

Additional Info

Field Value
Created December 2, 2024
Last update December 2, 2024
Defined In https://doi.org/10.48550/arXiv.2309.13770
Author Yichen Xu
More Authors
Zihan Xu
Wenhao Chai
Zhonghan Zhao
Enxin Song
Gaoang Wang
Homepage https://arxiv.org/abs/2304.14108