It is a monolingual transformers model pretrained on a large corpus of Hindi data (100GB+) in a self-supervised fashion.
