Datasets:
Dataset Viewer issue: DatasetWithScriptNotSupportedError
The dataset viewer is not working.
Error details:
Error code: DatasetWithScriptNotSupportedError
We would like to maintain dataset versioning via Github and avoid manual maintanence of future updates on two platforms.
Instead, we plan to make changes on Github primarily and use the metadata file as a handshake for versioning of datasets for cross-platform access (for eg on HuggingFace).
Please consider exception or suggest an alternative that allows automated way of version controlling datasets hosted on Github.
Thank you
👋 Before opening the discussion, have you considered removing the loading script and relying on automated data support?
You can use convert_to_parquet from the datasets library.
Hi ! Dataset scripts are legacy / not-recommended ways to define a dataset nowadays :/
I'd recommend having the dataset versioned directly here, or have a github-action to sync the HF repository regularly.
Having the files directly here lets AI builders
- use the dataset without having to
trust_remote_code(which is not allowed in some environments) - use the dataset with the libraries that have HF integrations like pandas, dask, spark or duckdb
It will also enable the Dataset Viewer for people to explore / query the dataset.
Let me know if I can help with anything !