BSC-LT/MULTI_corpus

Dataset Card for MULTI-Corpus Dataset Summary This corpus was compiled as part of the TRAIN project (Traducción Automática para la Inclusión, Automatic Translation for Inclusion), funded by MCIN/AEI and ERDF. It aggregates 15,191,441 sentence-level entries covering four extremely low-resource languages: Tamazight/Amazigh (ZGH), Pashto (PS), Wolof (WO), and Romani (ROM), paired with one or more high-resource counterparts (English, Spanish, French), plus monolingual… See the full description on the dataset page: https://huggingface.co/datasets/BSC-LT/MULTI_corpus.