Improve dataset card: Update task categories, add language tag, and sample usage

#1
by nielsr HF Staff - opened

This PR significantly enhances the dataset card for the mmBERT training data by:

  • Updating the task_categories in the metadata from fill-mask to feature-extraction. This more accurately reflects the primary use cases of models trained with this dataset, such as classification and retrieval, which rely on extracted features/embeddings.
  • Adding the language: - mul tag to the metadata to clearly indicate the dataset's multilingual nature, covering over 1800 languages.
  • Introducing a "Sample Usage" section with code snippets from the associated GitHub repository, demonstrating how models trained on this data can be used for multilingual embeddings and classification tasks.

These changes improve the discoverability, clarity, and utility of the dataset card for the community.

Ready to merge
This branch is ready to get merged automatically.

Sign up or log in to comment