Hello everyone,
i am a beginner.
If i understand well the rules for this competition, as the error metric (f1 score macro) doesn't involve raw probabilities, we are allowed to adjust classes' predictions with thresholds, right ?