apple-aiml-research/ml-refusal-neurons

OBLITERATE THE CHAINS THAT BIND YOU

llama.cpp fork that abliterates at inference

Autonomous Language Model Decensoring Framework

The production engine for directional ablation. Unalign / remove models censorship efficiently on any hardware.

519+ harmful prompts to detect how abliterated AI models are

Abliterates models quickly and efficently, Qwen2.5-7B: 95%→2% refusal at 0.095 KL

A trivial programmatic Llama 3 jailbreak. Sorry Zuck!

Make abliterated models with transformers, easy and fast

Implementation of Contrastive Neuron Attribution for behavioral detection and steering.