WIT: Wikipedia-based Image Text Dataset For Multimodal Multilingual Machine Learning

Abstract

The milestone improvements brought about by deep representation learning and pre-training techniques have led to large performance gains across downstream NLP, IR and Vision tasks. Multimodal modeling techniques aim to leverage large high-quality visio-linguistic datasets for learning complementary information (across image and text modalities). In this paper, we introduce the Wikipedia-based Image Text (WIT) Dataset (https://github.com/google-research-datasets/wit) to better facilitate multimodal, multilingual learning. WIT is composed of a curated set of 37.6 million entity rich image-text examples with 11.5 million unique images across 108 Wikipedia languages. Its size enables WIT to be used as a pretraining dataset for multimodal models, as we show when applied to downstream tasks such as image-text retrieval. WIT has four main and unique advantages. First, WIT is the largest multimodal dataset by the number of image-text examples by 3x (at the time of writing). Second, WIT is mass

Stats

• citations145
• S2 citations—
• github stars1107
• HF likes0
• heat score22.32
• arxiv keysrinivasan2021wit

Code

• google-research-datasets/wit—

Related papers

• Multilingual Diversity Improves Vision-language Representations (2024)2.26
• Entity Image And Mixed-modal Image Retrieval Datasets (2025)1.56
• Mr. Right: Multimodal Retrieval On Representation Of Image With Text (2022)0.00
• Imagebert: Cross-modal Pre-training With Large-scale Weak-supervised Image-text Data (2020)0.00
• Wikimute: A Web-sourced Dataset Of Semantic Descriptions For Music Audio (2023)5.24
• Atomic: An Image/text Retrieval Test Collection To Support Multimedia Content Creation (2023)9.02
• Image Search Using Multilingual Texts: A Cross-modal Learning Approach Between Image And Text (2019)0.00
• Self-supervised Visual Representations For Cross-modal Retrieval (2019)7.50