OpenGVLab

/

InternVL3_5-4B-Instruct

like 2

Follow

OpenGVLab 1.52k

This PR enhances the model card by adding more descriptive tags to the metadata. These tags better reflect the model's capabilities and underlying architecture, improving its discoverability on the Hugging Face Hub.

The added tags include:

• multimodal: The model is explicitly described as an "open-source multimodal model".
• vision-language-model: This is a more specific classification for the model type.
• reasoning: The paper highlights "reasoning capability" as a significant advancement.
• agentic: The model supports "GUI interaction and embodied agency" and performs well on "agentic tasks".
• multilingual: The existing metadata language: - multilingual and the GitHub README's "Multilingual Zero-Shot Image-Text Retrieval" section confirm its multilingual capabilities.
• qwen: The config.json file specifies Qwen3ForCausalLM as the language model component.

No other changes to the content or existing metadata were deemed necessary as they are already well-documented.