Measuring Cross-lingual Transfer in Bytes

arXiv:2404.08191v1 Announce Type: new
Abstract: Multilingual pretraining has been a successful solution to the challenges posed by the lack of resources for languages. These models can transfer knowledge to target languages with minimal or no examples. Recent research suggests that monolingual models also have a similar capability, but the mechanisms behind this transfer remain unclear. Some studies have explored factors like language contamination and syntactic similarity. An emerging line of research suggests that the representations learned by language models contain …

abstract arxiv capability challenges cross-lingual cs.cl examples knowledge languages measuring multilingual pretraining research resources solution transfer type