This PR significantly enhances the model card for **SPECS: Specificity-Enhanced CLIP-Score for Long Image Caption Evaluation** by:
* Adding the `pipeline_tag: zero-shot-image-classification` to the metadata, improving discoverability on the Hub (e.g., https://huggingface.co/models?pipeline_tag=zero-shot-image-classification).
* Including the paper title and a direct link to the Hugging Face paper page: [SPECS: Specificity-Enhanced CLIP-Score for Long Image Caption Evaluation](https://huggingface.co/papers/2509.03897).
* Adding a link to the official GitHub repository: https://github.com/mbzuai-nlp/SPECS.
* Incorporating the paper abstract into the model card content for better understanding.
* Adding a comprehensive BibTeX citation for proper academic attribution.
The existing Python usage example has been retained as it is directly applicable and sourced from the official GitHub repository. No `library_name` was added as there was no evidence of compatibility with Hugging …
