Guide-linked quiz · Medium Level
Label Bias and Annotator Disagreement Quiz
Examine subjective labels, annotator identity and instructions, proxy targets, disagreement, and safer ways to preserve uncertainty in training data.
Question 1 of 8
A model learns “toxic” labels that reflect annotators’ assumptions about dialect. What risk is most direct?
Keep testing yourself
More quizzes picked for your level.