REPA-E

/

iREPA-collections

like 3

Follow

REPA-E 16

This PR improves the model card for the iREPA model by adding key metadata and essential information:

• Added the pipeline_tag: unconditional-image-generation to correctly classify the model on the Hub.
• Specified library_name: diffusers as the model uses components and training practices compatible with the Diffusers library (e.g., sdvae-ft-mse-f8d4 VAE, accelerate launch). This enables the automated "how to use" widget.
• Linked to the official paper: What matters for Representation Alignment: Global Information or Spatial Structure?.
• Included links to the project page (https://end2end-diffusion.github.io/irepa) and the GitHub repository (https://github.com/end2end-diffusion/irepa).
• Provided a clear summary of the paper's findings and the iREPA method.
• Added the core iREPA code snippet directly from the GitHub README as a sample usage example.
• Included the primary visual from the project page for better presentation.
• Added the BibTeX citation for proper attribution.

Please review and merge this PR if everything looks good.