Your question is Toxicity Training with Class Imbalance. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
How do you handle severe class imbalance when training toxicity classifiers on short-form text and comment data?
Discuss data splitting, text representation, sampling or loss weighting, threshold selection, and evaluation metrics. Include how you would avoid leakage, preserve minority examples, and monitor performance across relevant toxicity categories and user populations.