Hi! I'm Niels from the Hugging Face community science team.
This PR improves the dataset card by:
- Adding the `text-generation` task category to the metadata.
- Adding the `license: other` tag to the metadata (as the dataset aggregates multiple sources).
- Adding the `arxiv` paper ID to the metadata.
- Correcting the `default` configuration name and the usage code snippet to match the actual configuration name (`minhash_deduped`) defined in the YAML.