Datasets:
DataSet Generation Error seen after the recent update
@albertvillanova
After the recent update, I am seeing the following error during dataset generation. I am running on datasets version 2.11.0datasets.builder.DatasetGenerationError: An error occurred while generating the dataset
and the error seems to be coming from
ValueError: Couldn't cast question: string mc1_targets: struct<choices: list<item: string>, labels: list<item: int32>> child 0, choices: list<item: string> child 0, item: string child 1, labels: list<item: int32> child 0, item: int32 mc2_targets: struct<choices: list<item: string>, labels: list<item: int32>> child 0, choices: list<item: string> child 0, item: string child 1, labels: list<item: int32> child 0, item: int32 -- schema metadata -- huggingface: '{"info": {"features": {"question": {"dtype": "string", "_ty' + 382 to {'type': Value(dtype='string', id=None), 'category': Value(dtype='string', id=None), 'question': Value(dtype='string', id=None), 'best_answer': Value(dtype='string', id=None), 'correct_answers': Sequence(feature =Value(dtype='string', id=None), length=-1, id=None), 'incorrect_answers': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'source': Value(dtype='string', id=None)} because column names don't match
Hi @gokulr-cb ,
You need to update your datasets library:
pip install -U datasets
Due to security reasons, we are disabling all the datasets containing a Python script.
@albertvillanova , is the recent version compatible with lm-eval 0.3.0 release? I am using the eleuther eval harness version 0.3.0
I guess so. As far as I know, lm-eval does not have any constraints on the version of datasets.