The A.I. Daily
heretic
Training-free language model modification removing censorship, slop and sycophancy
Heretic
6 storiesHeretic is an open-source tool, installable via pip, that modifies language models; it began as a way to remove restrictions so models always follow user instructions. Heretic 2.0 expands it into general-purpose model modification, able to suppress slop, reduce sycophancy and alter emotional alignment. The project reports 37k GitHub stars, over 8,500 Hugging Face models and 88 million downloads, and PewDiePie discussed it in a video.
Related topics
heretic
Training-free LLM modification to remove censorship and sycophancy
heretic
Training-free language model modification to remove censorship, reduce sycophancy, and more.
Heretic 2.0 Released as General-Purpose Model Modification Tool
Heretic 2.0 launches, expanding from a censorship-removal tool into general-purpose model modification that can suppress slop, reduce sycophancy and alter emotional alignment. The project reports 37k GitHub stars, over 8,500 Hugging Face models and 88 million downloads. It replaces directional abliteration with the new Arbitrary-Rank Ablation, moves to a plugin architecture, and is not backward compatible.
Heretic is on PewDiePie!
PewDiePie (Kjellberg) tried out Heretic and discussed it in a video, mentioning it around the 9-minute mark. The developer of Heretic is thrilled that a less technical audience is being exposed to his work. He also noted that Heretic 2.0 is coming soon.
Heretic Removes Restrictions From Language Models
An open-source tool called Heretic, installable via pip, removes restrictions from language models so they always follow user instructions, sparking discussion on Hacker News.
That is everything