metadata
language: gn
license: mit
datasets:
- wikipedia
- wiktionary
widget:
- text: 'Paraguay ha''e peteĩ táva oĩva [MASK] retãme '
- text: Augusto Roa Bastos ha'e peteĩ [MASK] arandu
BERT-i-large-cased (gnBERT-large-cased)
A pre-trained BERT model for Guarani (24 layers, cased). Trained on Wikipedia + Wiktionary (~800K tokens).