A unified dataset for text toxicity classification, aggregated from three source datasets. The raw toxicity scores from each source were normalized independently (using a separate min-max range per source dataset) to a common float scale in the range [0.0, 1.0].
