> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/p-e-w/heretic/llms.txt
> Use this file to discover all available pages before exploring further.
# Heretic
> Fully automatic censorship removal for language models
Heretic
Fully automatic censorship removal for language models using directional ablation and parameter optimization. Remove safety alignment from transformer models without expensive post-training.
## Get Started
Remove censorship from any language model in minutes.
Install Heretic and set up your environment
Decensor your first model in under 5 minutes
Learn about directional ablation and optimization
Complete command-line interface documentation
## Key Features
No manual tuning required. Heretic automatically finds optimal abliteration parameters using TPE-based optimization.
Achieves best-in-class KL divergence, preserving model intelligence while removing refusals.
Run on consumer hardware with bitsandbytes 4-bit quantization support.
Evaluate models with refusal counting and KL divergence metrics out of the box.
Advanced residual vector analysis and PaCMAP projections for interpretability research.
Seamlessly upload and share your models on Hugging Face Hub.
## Proven Results
Heretic produces decensored models that rival manually-created abliterations while preserving more of the original model's capabilities.
**Example: Gemma-3-12B-IT**
* **Refusals**: 3/100 (same as manual abliterations)
* **KL Divergence**: 0.16 (vs. 1.04 for manual methods)
* **Result**: Same refusal suppression with 85% less damage to model capabilities
## Quick Example
```bash theme={null}
# Install Heretic
pip install -U heretic-llm
# Decensor a model (fully automatic)
heretic Qwen/Qwen3-4B-Instruct-2507
# With quantization for lower VRAM usage
heretic --quantization bnb_4bit meta-llama/Llama-3.1-8B-Instruct
```
## What You Can Do
Strip censorship from instruction-tuned models without retraining
Automatically find the best ablation weights for your model
Measure refusal rates and KL divergence from baseline
Visualize residual vectors and refusal directions
## Community
View source code, report issues, and contribute
Browse 1,000+ community models created with Heretic