YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
AfroXLMR fine tuned for Kinyarwanda news topic classification
This model classifies Kinyarwanda news articles into 14 topics. It is Davlan/afro-xlmr-base fine tuned on the cleaned KINNEWS dataset (Niyongabo et al., 2020).
Test results on the cleaned test set (2027 articles): accuracy 0.7834, macro F1 0.6666.
Settings: 3 epochs, maximum length 256 tokens, learning rate 2e-5, batch size 16.
Limits: articles are cut at the maximum length, small topics such as history and environment are predicted less reliably, and the model was trained on news text only. See the original model card of the pretrained model for its license terms.
- Downloads last month
- 28
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support