p-e-w/heretic

Training-free language model modification: Remove censorship, suppress slop, reduce sycophancy, and more

745 best points · 380 comments · 1 Hacker News post · 39,797 GitHub stars · View on GitHub

Hacker News discussions

• Heretic: Automatic censorship removal for language models (745 points, 380 comments, by melded, 2025-11-16)