ShaoRun

/

RS-EoT-7B

like 3

This PR significantly enhances the model card for RS-EoT-7B by:

• Adding pipeline_tag: image-text-to-text to the metadata for improved discoverability on the Hugging Face Hub, categorizing it as a model that processes images and outputs text.
• Adding library_name: transformers to the metadata. This enables the automatic display of the "How to use" widget with transformers code snippets, greatly improving user experience and model accessibility, as demonstrated by the provided usage examples.
• Including the "Dataset Loading" section from the original GitHub README, offering immediate guidance on how to access and utilize the associated dataset.
• Updating the image URL for the "RS-EoT Overview" figure to a raw GitHub link (https://github.com/GeoX-Lab/Asking_like_Socrates/blob/main/assets/Fig1.jpg?raw=true) to ensure it renders correctly on the Hub.
• Adding the "Code Structure" section from the original GitHub README, providing valuable context about the project's organization and different components (SocraticAgent, LLaMA-Factory, EasyR1).
• Updating the "Citation" section with the full BibTeX entry from the arXiv paper, allowing for proper attribution.
• Adjusting the heading levels for VQA and Visual Grounding examples within "Quick Start" to #### for better semantic structure.

These changes make the model card more comprehensive, functional, and user-friendly.