作者位次 #5/6ACL 2026Poster· Natural Language Processing
A frique LLM : How Data Mixing and Model Architecture Impact Continued Pre-training for A frican Languages
Hao Yu · Tianyi Xu · Michael A. Hedderich · Wassim Hamidouche · Syed Waqas Zamir · David Ifeoluwa Adelani
摘要
Large language models (LLMs) are increasingly multilingual, yet open models continue to underperform relative to proprietary systems, with the gap most pronounced for African languages. Continued pre-training (CPT) offers a practical route to language adaptation, but improvements on demanding cap...