datasets 2.18.0

HuggingFace community-driven open-source library of datasets

Stars: 18327, Watchers: 18327, Forks: 2497, Open Issues: 673

The huggingface/datasets repo was created 4 years ago and the last code push was 2 hours ago.
The project is extremely popular with a mindblowing 18327 github stars!

How to Install datasets

You can install datasets using pip

pip install datasets

or add it to a project with poetry

poetry add datasets

Package Details

Author

HuggingFace Inc.

License

Apache 2.0

Homepage

https://github.com/huggingface/datasets

PyPi:

https://pypi.org/project/datasets/

GitHub Repo:

https://github.com/huggingface/datasets

Classifiers

• Scientific/Engineering/Artificial Intelligence

Related Packages

Errors

A list of common datasets errors.

Code Examples

Here are some datasets code examples and snippets.

GitHub Issues

The datasets package has 673 open issues on GitHub

• AutoTokenizer hash value got change after datasets.map
• [TypeError: Couldn't cast array of type] Cannot load dataset in v1.18
• Make ted_talks_iwslt dataset streamable
• Dataset.shuffle(seed=None) gives fixed row permutation
• Adding CC-100: Monolingual Datasets from Web Crawl Data (Datasets links are invalid)
• Labels conflict when loading a local CSV file.
• DuplicatedKeysError of NewsQA dataset
• Dataset Card Creator drops information for "Additional Information" Section
• Fix host URL in The Pile datasets
• The Pile cannot connect to host
• Add a metadata field for when source data was produced
• Extend support for streaming datasets that use os.path.relpath
• Extend support for streaming datasets that use os.path.relpath
• Consider adding ipywidgets as a dependency.
• Add Fon language tag

See more issues on GitHub