Kangheng

/

OVR-7B-ColdStart

like 2

This PR significantly enhances the model card by:

• Adding the pipeline_tag: image-text-to-text to the metadata, ensuring the model is discoverable under this pipeline on the Hub.
• Adding library_name: transformers to the metadata, indicating its compatibility with the Hugging Face Transformers library.
• Updating and explicitly adding links to the paper (Open Vision Reasoner: Transferring Linguistic Cognitive Behavior for Visual Reasoning), the project page (https://weiyana.github.io/Open-Vision-Reasoner/), and the GitHub repository (https://github.com/linkangheng/Open-Vision-Reasoner) for better visibility and access.
• Integrating more detailed sections from the original GitHub README, including:
  • Expanded "Performance Results" with sub-sections for Language, Visual, and Cognitive Behavior Analysis.
  • A new "Training Pipeline" section to describe the two-stage training paradigm.
  • A "Roadmap" section for future plans.
  • The "Citation" information for proper attribution.