heretic: Fully automatic censorship removal for language models

Heretic is a fully automated tool that removes safety restrictions from language models without requiring expensive retraining or technical expertise. The system uses advanced directional ablation techniques combined with parameter optimization to automatically find the best settings for removing censorship while preserving the model's original capabilities. Testing shows Heretic-modified models achieve the same low refusal rates as manually created versions but with significantly less damage to core intelligence, requiring only about 45 minutes to process models like Llama-3.1-8B.