This dataset can be used to build extractive-QA and Language Models. It is an aggregation and balancing of 2 previous datasets: VilaQuAD and ViquiQuAD.
Splits have been balanced by kind of question, and unlike other datasets like SQuAD, it only contains, per record, one question and one answer for each context, although the contexts can repeat multiple times.
This dataset was developed by BSC TeMU as part of Projecte AINA, to… See the full description on the dataset page: https://huggingface.co/datasets/projecte-aina/catalanqa.
