IMPORTANT: the label column in the test set has dummy values as these were not provided (see original readme section for explanation)
DatasetDict({
train: Dataset({
features: ['claimID', 'claim', 'label', 'claimURL', 'reason', 'categories', 'speaker', 'checker'… See the full description on the dataset page:
https://huggingface.co/datasets/pszemraj/multi_fc.