Datasets:

spacemanidol

/

cc-stories

like 9

Change _DATASET_URLS from 'all' to 'train'.
'all' is a reserved keyword corresponding to the union of all splits, so cannot be used as key in "._split_generator()."
https://github.com/huggingface/datasets/blob/main/src/datasets/builder.py#L975

I tried to find a way to ignore splits but bumped into this thread.
https://discuss.huggingface.co/t/dataset-with-no-splits/8515/2