markendo

/

llava-extract-from-scratch-qwen3-0.6B

like 0

This PR significantly enhances the model card by:

• Adding pipeline_tag: image-text-to-text, making the model discoverable under this pipeline on the Hugging Face Hub.
• Specifying library_name: transformers as the model's config.json indicates compatibility with the Transformers library, enabling the automated "how to use" widget.
• Including direct links to the paper (Downscaling Intelligence: Exploring Perception and Reasoning Bottlenecks in Small Multimodal Models), the project page, and the GitHub repository for easy access to resources.
• Incorporating a "Usage" section with clear bash code snippets directly from the original GitHub README to demonstrate model evaluation.
• Adding the BibTeX "Citation" section from the GitHub README.

Existing content, including the detailed "File information", has been retained as part of the current model card's structure.