Error when loading the dataset.
#1
by
avacaondata
- opened
I tried to load the dataset with the last version of HF datasets library:
wnli = load_dataset("PlanTL-GOB-ES/wnli-es")
But throws an error:
ValueError: Couldn't cast
index: int64
sentence1: string
sentence2: string
-- schema metadata --
pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 603
to
{'index': Value(dtype='int64', id=None), 'sentence1': Value(dtype='string', id=None), 'sentence2': Value(dtype='string', id=None), 'label': Value(dtype='int64', id=None)}
because column names don't match
Thank you for reporting this error
It seems that the error was coming from the test set, we added the label column from the test set with -1 (instead of adding it in the data loader) and now it seems to work.
Also, the name of the data loader (.py) script was different from the dataset name (it was wnli.py instead of wnli-es.py).
Hope this solves the problem