xlm-roberta wolof
Wolof-tuned encoders
Official site consultedAI: models
Last verified: 1 October 2026
XLM-RoBERTa-base model fine-tuned on Wolof for fill-mask, using the Bible (OT), the OPUS corpus and articles from Lu Defu Waxu, Saabal and Wolof Online. It scores 68.31 F1 on MasakhaNER versus 63.86 for the base model. The page consulted covers only the base version.
Services
- Wolof fill-mask
- Named entity recognition (68.31 F1)
Access requirements
Weights on Hugging Face; no license stated on the page.
Sources
- xlm-roberta wolof (base, large) (new tab), verified on 1 October 2026Official site consulted