Thats the thing about a normalization system, it is going to normalize outputs because its not built to output uniqueness, its to winnow uniqueness to a baseline. That is good in some instances, assuming that baseline is correct, but it also closes the aperture of human expression.
Token selection is based off normalization, even if you train a model to produce outlier answers, even in that process you are biasing to a subset of outliers, which is inherently normalizing.
