JoeLeelyf

/

Skyra-RL

like 1

This PR adds a comprehensive model card for the Skyra model.

It includes:

• Adding the license: cc-by-4.0 as stated in the GitHub repository.
• Adding the library_name: transformers due to clear evidence in config.json (Qwen2_5_VLForConditionalGeneration, transformers_version).
• Setting the pipeline_tag: video-text-to-text to accurately reflect the model's functionality of processing video for text-based detection and explanation.
• Linking to the research paper (Skyra: AI-Generated Video Detection via Grounded Artifact Reasoning), project page (https://joeleelyf.github.io/Skyra/), and GitHub repository (https://github.com/JoeLeelyf/Skyra).
• Incorporating key sections from the paper's GitHub README (abstract, introduction, capabilities, dataset, methodology, results, usage, license, and citation) to provide a complete overview.
• The "Usage" section is copied directly from the GitHub README, as no specific transformers inference code snippet was present in the original README for this model, aligning with the instruction not to make up code.

Please review and merge this PR.