nmndeep

/

CLIC-ViT-B-16-224-CogVLM

like 0

This PR significantly enhances the model card for CLIC-ViT-B-16-224-CogVLM by adding comprehensive information and improving its usability:

• Metadata: Added key YAML metadata:
  • pipeline_tag: zero-shot-image-classification to categorize the model accurately.
  • library_name: open_clip to enable the automated "How to use" widget on the Hub.
  • license: apache-2.0 for clarity on usage rights.
  • datasets: Linked to the nmndeep/CLIC-CogVLM-relabelled-Laion and tomg-group-umd/pixelprose datasets used for fine-tuning.
  • language: en for better discoverability.
• Documentation Links: Included direct links to the paper (Advancing Compositional Awareness in CLIP with Efficient Fine-Tuning), the project page (https://clic-compositional-clip.github.io/), and the GitHub repository (https://github.com/AmitPeleg/CLIC).
• Abstract: Added the paper's abstract to provide immediate context about the model's capabilities and purpose.
• Sample Usage Fix: Corrected the OpenCLIP usage example by adding from urllib.request import urlopen to make the code snippet fully runnable.
• Citation: Included the BibTeX citation provided in the GitHub repository.

These changes will improve the model's discoverability, provide clearer usage instructions, and offer more comprehensive background information for users.