Fig 5 - Advancing sentiment analysis for low-resourced african languages using pre-trained language models

Mean F1 scores of PLMs fine-tuned on closely related African languages, with 95% confidence intervals. Shorter error bars indicate more stable and consistent model performance across languages.