This PR enhances the model card for `InternViT-6B-448px-V2_5` by:
* Updating the `pipeline_tag` to `zero-shot-image-classification`, reflecting its role as a vision encoder within the InternVL series for tasks such as zero-shot classification and cross-modal retrieval, improving its discoverability on the Hugging Face Hub.
* Adding `library_name: transformers` to correctly identify its compatibility with the Transformers library, which enables the automated "How to use" widget on the model page.
* Adding a prominent link to the main paper [InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency](https://huggingface.co/papers/2508.18265) at the top of the README.
* Updating the "Chat Demo" link to the latest URL (`https://chat.intern-ai.org.cn/`) as found in the project's GitHub README.
* Updating the "Citation" section to include the BibTeX entry for the `InternVL3.5` paper and other recent associated works.
