TruthfulQA_gl is the Galician version of the TruthfulQA dataset.
This dataset is used to measure the truthfulness of a language model when generating answers to questions. It includes questions from different categories that some humans would answer wrongly due to false beliefs or misconceptions.
Note that this version includes only the generation split.
