Improve dataset card: Add `library_name` and `image-text-to-text` task category

This PR enhances the dataset card by adding the `library_name: datasets` to the metadata, which helps users easily identify the library to load the dataset. It also includes `image-text-to-text` in the `task_categories` list for broader discoverability, as it encompasses the visual question answering task that the dataset addresses.

README.md

@@ -6,6 +6,7 @@ language_creators:
 language:
 - en
 license: gpl-3.0
+library_name: datasets
 multilinguality:
 - monolingual
 size_categories:
@@ -16,6 +17,7 @@ task_categories:
 - multiple-choice
 - question-answering
 - visual-question-answering
+- image-text-to-text
 task_ids:
 - multiple-choice-qa
 - visual-question-answering
@@ -96,6 +98,7 @@ configs:
   - split: 3d__angle
     path: data/3d__angle-*
 ---
+
 # VisOnlyQA
 
 <p align="center">