categoryScores

Per-category raw scores output by the model, denoting the model's confidence that the input violates the OpenAI's policy for the category. The value is between 0 and 1, where higher values denote higher confidence. The scores should not be interpreted as probabilities.