internlm

/

Spark-VL-7B

like 10

Follow

Intern Large Models 814

This PR enhances the model card for Spark-VL-7B by adding crucial metadata and completing the citation:

• pipeline_tag: video-text-to-text: Improves discoverability on the Hugging Face Hub for multimodal models that generate text from both image and video inputs, reflecting the model's capabilities as an LVLM.
• library_name: transformers: Enables the automated "How to use" widget, as the model is fully compatible with the transformers library, evidenced by the provided sample usage.
• tags: [lvlm, reasoning, multimodal, qwen]: Adds additional descriptive tags for better searchability and categorization, aligned with the model's architecture (Qwen base) and capabilities (LVLM, reasoning, multimodal).
• Citation: Populates the Citation section with the correct BibTeX entry from the associated paper, replacing the "TBD" placeholder.

These changes will make the model more easily discoverable and ensure its information is consistent and accurate.