youngsheen

/

SimVQ

like 3

This PR significantly enhances the model card for SimVQ by adding essential information and improving its discoverability on the Hugging Face Hub.

Key updates include:

• Metadata: Added pipeline_tag: image-to-image to categorize the model for relevant tasks. The existing license: mit is retained.
• Model Description: Incorporated a summary of the model from the paper's abstract, explaining its purpose and approach.
• Links: Provided direct links to the scientific paper (Addressing Representation Collapse in Vector Quantized Models with One Linear Layer) and the official GitHub repository (https://github.com/youngsheen/SimVQ).
• Key Results: Integrated crucial visual and quantitative data from the GitHub README, including the SimVQ algorithm diagram, detailed quantitative comparison tables for image and audio reconstruction, and visualization examples of reconstructed images and audio. All relative image paths have been updated to absolute raw GitHub URLs.
• Acknowledgements: Included the acknowledgements section from the original repository.
• Citation: Added a BibTeX entry for easy academic citation.

This update ensures the model card is informative, well-linked, and accurately represented for the community.