Insulter uses Tensorflow to generate insults based on a large dataset I found on Kaggle (this was the ultimate weekend project).
Here's the type of insults it generates:
Fucker you are a shit on the back of the post of the posts are a shit on the same sex and the truth the fact that is a fucking like a stupid and the truth because they are a shit on the post of the post of the problem is a fucking like a stupid and the fact that you are a shit on the same sex and the post of the political companies who are a shit on the political computer and show the taxes and the problem is a fucking like a shit on the time to be a fucking like a shit on the same sex and the post sex and stupid one of the same sex and the fact that is a fucking like a stupid and the taxes and stupid ass of the people that is a fucking like a stupid and stupid ass of your bastards should be a fucking like a stupid and stupid ass of the back to be a fucking like a shit on the team and the facts that is a fucking like a stupid and the fuck up the party of the problem is a fucking like a stupid and the taxes and stupid ass of the more particular something with the most state the team and the
I desire to bring insulter to its second version using GPT-2 or GPT-3.