mopatik/setswana-offensive-977
Domaine:
natural language processingType de record:
dataset
Créateur:
mop
Hôte:
Tasks: Binary offensive language detection (0 = Non-offensive, 1 = Offensive)Language: Setswana (tsn; Botswana/South Africa/Namibia)Size: 977 instances (477 offensive; 500 non-offensive)
Protocol: 80/20 holdout (tag-free, unaugmented). 5-fold CV on the 80% train.
Evaluation: Holdout tag-free, mirrors deployment.
Labels: {0: Non-offensive, 1: Offensive}