This PR improves the dataset card by:
- Adding `task_categories: ['video-text-to-text']` to the metadata, enabling better discoverability on the Hugging Face Hub under relevant AI tasks.
- Adding `language: ['en']` as the dataset content is in English.
- Including relevant `tags` such as `autonomous-driving`, `nuscenes`, `temporal-reasoning`, `video-question-answering`, and `vlm` for more detailed categorization.
- Adding a "Sample Usage" section with a Python code snippet directly from the GitHub README, making it easier for users to get started with the dataset. This section also includes important setup information for loading the benchmark splits.
