#natural language process...
Jul 2026
Expanding the Lexicon of Ge'ez Based African Languages: A Comparative Study of Amharic and Tigrinya
VEXMLM demonstrates that vocabulary expansion and tokenizer adaptation and integration provide an effective, computationally efficient path to improve multilingual models for underrepresented languages without retraining from scratch.
Hailay Kidu Teklehaymanot, Debela Desalegn Yadeta, Wolfgang Nejdl
· arXiv.org · 0 citations