I just made two different dittos both with local toxicity filtering ideas.
1. Uses a local model running in LM Studio (with cors enabled). It prompts the LLM to get a toxicity level for each post. Could use a better prompt...
https://25yhg55w2nqfk006nhvvejnvprnxomaoqd0vv7035n1xoo5001ditto-aimod.nsite.lol/trends
2. This uses a tiny bert model that is stored in your browser. It's pretty laggy and the analysis isn't great but it's a proof of concept lol.
Just ideas. I'm considering using the 1st method. With a better prompt it could be pretty good. #filters #moderation #nostrdev

Ditto
Ditto
Your content. Your vibe. Your rules.


