I’m using Google’s perspective api, it’s quite nuanced in its assessment of different types of abusive content. I set a threshold on 2 parameters that I feel do

I’m using Google’s perspective api, it’s quite nuanced in its assessment of different types of abusive content. I set a threshold on 2 parameters that I feel does a great job of filtering out incendiary comments but allowing jokes that might be slightly risky but still acceptable. That’s a great point about the iOS local inference option, I’m currently using Expo so it’s cross platform and my current set up is almost instant with Perspective api. I will def explore using Swift and apples native ML models for future apps though so thanks for suggesting