# Multilingual zero-shot learning

BLOOMZ and mT0 are a series of multilingual instruction-following models. By fine-tuning pre-trained models on a cross-lingual task mixed dataset, they have zero-shot cross-lingual generalization ability.

XGLM-7.5B is a multilingual autoregressive language model with 7.5 billion parameters, supporting 30+ languages, trained on a diverse corpus of 500 billion subword tokens.