Add pipeline tag and library name

This PR adds a pipeline tag and library name to improve model discoverability and usability. The pipeline tag `image-text-to-text` ensures the model is easily found by users interested in this specific task. The `transformers` library name indicates the intended integration with the Transformers library.

README.md

@@ -5,6 +5,8 @@ tags:
 - vision-language-model
 - contrastive learning
 - self-supervised learning
+library_name: transformers
+pipeline_tag: image-text-to-text
 ---
 
 **COSMOS Model**