Improve dataset card: Add task categories, paper abstract, and model sample usage
#1
by nielsr HF Staff - opened
This PR improves the dataset card for the mmBERT Training Data by:
- Updating the
task_categoriesmetadata to includefeature-extraction, which better reflects the primary use cases (classification, retrieval) of models trained with this data, alongside the existingfill-maskpre-training objective. - Adding
language: - multo the metadata, explicitly tagging the dataset as multilingual. - Adding explicit links to the paper (Hugging Face Papers page) and GitHub repository in the main content for better visibility.
- Including the full Paper Abstract to provide comprehensive context about the dataset's purpose and the models it supports.
- Adding a Sample Usage section with code snippets from the GitHub README. These examples demonstrate how to use
mmBERTmodels (trained using this dataset) for tasks like installation, obtaining multilingual embeddings, and masked language modeling, making the dataset card more practical and informative.