Hi! This PR improves the model card by:
- Adding the `video-text-to-text` pipeline tag to the metadata for better categorization.
- Including a "Sample Usage" section with a code snippet from the official repository to assist users with inference.
- Ensuring all relevant links (GitHub and Paper) are clearly presented.
