Datasets:

---

JoeLeelyf

/

ViF-Bench

like 1

This PR adds a comprehensive dataset card for the ViF-CoT-4K dataset.

The changes include:

• Adding task_categories: ['video-text-to-text'], license: 'cc-by-4.0', language: 'en', and relevant tags to the metadata.
• Linking to the associated paper: Skyra: AI-Generated Video Detection via Grounded Artifact Reasoning.
• Linking to the project page: https://joeleelyf.github.io/Skyra/.
• Linking to the GitHub repository: https://github.com/JoeLeelyf/Skyra.
• Including a detailed introduction and description of the ViF-CoT-4K dataset, its scale, annotation, and purpose, along with illustrative images.
• Providing sample usage instructions and code snippets for data preparation, supervised fine-tuning (SFT), reinforcement learning (RL), and evaluation, directly extracted from the GitHub README.
• Adding the BibTeX citation for the paper.
• Explicitly stating the license terms.

This update significantly improves the discoverability and usability of the ViF-CoT-4K dataset for the community.