Text-Curation

Label column Feature columnsOptional numeric features. Pred prob columnsOptional. One per class. Max issue rows Run Datalab

Label column Feature columns CV folds Seed Run CleanLearning

Annotator columnsSelect 2+ columns, one per annotator. Pred prob columnsOptional. One column per class probability. Consensus methodbest_qualitymajority_vote Run Multi-Annotator

Label column Pred prob columns Score methodself_confidencenormalized_marginconfidence_weighted_entropy Adjust for class imbalance Top rows Run Rank Scoring

Label column Pred prob columns Get Health Report

Label column Pred prob columns Compute Counts

Label column Pred prob columns Filter methodprune_by_noise_rateprune_by_classboth Frac noise Rank bynoneself_confidencenormalized_marginconfidence_weighted_entropy Run Filter

Feature columnsNumeric feature columns for KNN-based OOD. Pred prob columnsOptional. For entropy-based OOD. Label columnOptional. k (neighbors) Detect Outliers

Label column Feature columns k (neighbors) Compute Shapley Values