kumuji

/

Sa2VA-i-8B

like 0

This PR significantly enhances the model card for Sa2VA-i by:

• Updating Metadata: Correcting the pipeline_tag from image-text-to-text to image-segmentation to accurately reflect its capabilities in language-guided dense grounding and video object segmentation.
• Adding Paper Link: Linking to the official Hugging Face paper page for easy access to the research.
• Adding GitHub Link: Providing a direct link to the associated GitHub repository for code access.
• Enriching Content: Incorporating a comprehensive overview from the paper abstract and the GitHub README, including:
  • Authors and affiliations
  • A teaser image
  • Key improvements and detailed explanations
  • Performance highlights and competition results
  • A model zoo with links to other Sa2VA-i models
  • Comprehensive citation information for both Sa2VA-i and the original Sa2VA.

These changes provide a much more informative and accessible model card for the Hugging Face community.