This repository provides a European Portuguese (pt-PT) translation of HellaSwag, an adversarial benchmark for grounded commonsense inference. Given a short scenario, the model must choose the most plausible continuation from four options.Load with:
import datasets
data = datasets.load_dataset("ruibrogandrade/HellaSwag_PT-PT")
