Are there any documented thresholds that we can reference for content moderation using omni-moderation-latest? I would like to flag and remove pornography in my app, but when I test nsfw images they get “sexual” scores of 0.01 or lower. I worry that if I flag content at such low thresholds I will also be removing content that isn’t sexual.
It would be nice if there was a scale for each category that we could reference for moderation. If anyone has any numbers, please share.