Document Domain Dataset Sampling And Filtering

via “interactive web-based dataset exploration and subset creation”

5.85 billion image-text pairs foundational for image generation.

Unique: Web-based interface enables interactive exploration and subset creation without downloading billions of pairs; search demo provides immediate feedback on dataset content and filtering strategies

vs others: Lower barrier to entry than command-line or API-based access; however, web interface is likely slower and less flexible than programmatic access for large-scale filtering