Wikipedia Simple English Image and Text Data

Image and text data from Wikipedia Simple English, with crowdsourced annotations. The dataset size is categorized as between 100K and 1M instances, and it is stored in Parquet format.

Use Cases

• Train an image-to-text model using the image modality and associated text descriptions.
• Analyze crowdsourced annotations for image content across the monolingual English corpus.
• Build a multimodal dataset for tasks like image captioning by linking image and text modalities.

Strengths

• Dataset size is categorized as between 100,000 and 1,000,000 instances.
• Data includes both image and text modalities for multimodal tasks.
• Stored in the efficient Parquet file format.

Limitations

• Specific row count, column structure, and data fields are unknown.
• Geographic coverage is limited to the 'us' region tag, indicating potential bias.
• Crowdsourced annotations may introduce label noise or inconsistency.

Provenance

Source

Wikipedia Simple English

Collection Method

Crowdsourced annotations.

Freshness

Last updated on 2025-10-18.

Geography

United States (based on the 'us' region tag).