Datasets:

---

DogNeverSleep

/

RealUnify

like 2

This PR adds a comprehensive dataset card for the RealUnify benchmark, ensuring it is properly documented and discoverable on the Hugging Face Hub.

Key additions include:

• A link to the paper (RealUnify: Do Unified Models Truly Benefit from Unification? A Comprehensive Benchmark) and the GitHub repository (https://github.com/FrankYang-17/RealUnify).
• The explicit paper abstract for clear context.
• A detailed "Evaluation Pipeline" section from the GitHub README as sample usage.
• Appropriate task_categories (image-text-to-text, text-to-image), language: en, and relevant tags (multimodal, benchmark, unified-models).
• The custom dataset license and BibTeX citation.
• Visuals from the GitHub repository, with links updated to raw GitHub URLs for reliable rendering.