Cookies are used on the Lanfrica website to ensure you get the best experience.
This repository contains code to reproduce Better Quality Pre-training Data and T5 Models for African Languages which appears in the 2023 conference on Empirical Methods in Natural Language Processing (EMNLP).
AfriTeVa V2 was trained on 20 languages (16 African Languages + Arabic, English, French, Portuguese) as the successor to AfriTeVa and evaluated on text classification, summarisation, reading comprehension and machine translation.