Cultivar is a locale-oriented evaluation benchmark for multilingual machine translation. The name "Cultivar-flores" highlights that the benchmark is a cultivated variety of FLORES through localisation.
Imagine you test translation systems or LLMs on Spanish-to-English translation, this input is not your true use case: "La facultad de medicina de la Universidad de Stanford ..." (the Stanford University School of Medicine ...).
Furthermore, since the input… See the full description on the dataset page:
https://huggingface.co/datasets/pinzhenchen/Cultivar-flores.
