@sdethux

Digging into LLM training for more realistic “synthetic humans,” I found “uncensored” models (Dolphin 3.0, Hermes 3, LLaMA‑3.2). Some use ablation to reduce overly strict filters that block harmless prompts. Not for illegal use, just clearer answers + freer convo. Try Ollama: https://is.gd/AVe5Ah Models: https://is.gd/s21jlg https://is.gd/zzxwhD More reading: https://is.gd/higD9i

1 reply

0 recast

2 reactions