Study examines sample-size scaling of NLI on 16 African languages

Arxiv CS.CL (NLP/language models)

The paper systematically studies how increasing annotation data affects NLI performance on 16 African languages. Results show that performance improves with sample size, but gains vary significantly across languages and linguistic families.

Arxiv CS.CL (NLP/language models)·Anuj TiwariOluwapelumi OgunremuTerry Oko-odionJesujuwon EgbewaleHannah Nwokocha

8 days ago