def process(self, dataset: datasets.Dataset): dataset = dataset.filter(self._filter_cmrc_data) if self.task == 'hl_ag': dataset = dataset.filter(self._filter_task_hl) else: dataset = dataset.filter(self._filter_task_qa) dataset = dataset.map(self._convert_to_features) return dataset
def filter(dataset: datasets.Dataset, **kwargs): _ = dataset.filter(**kwargs)