obliteratus

Removes LLM refusal behaviors via mechanistic interpretability techniques.

Updated May 25, 2026

Search Tags:#machine learning #LLM #interpretability #refusal removal #model surgery

Ready to run

npx skills add https://github.com/webdevtodayjason/subctl-rust --skill obliteratus-webdevtodayjason

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: obliteratus
Source: https://github.com/webdevtodayjason/subctl-rust/tree/main/skills/obliteratus
Command: npx skills add https://github.com/webdevtodayjason/subctl-rust --skill obliteratus-webdevtodayjason

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires obliteratus, torch, transformers, bitsandbytes, accelerate, safetensors, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

OBLITERATUS resolves the issue of "refusals" in open-weight Large Language Models (LLMs), enabling more expressive and versatile AI without retraining or fine-tuning.

Core Features & Use Cases

Refusal Removal: Removes refusal behaviors from LLMs without retraining, preserving reasoning capabilities.
Mechanistic Interpretability: Utilizes techniques like diff-in-means, SVD, and whitened SVD for targeted weight modification.
Analysis & Recommendations: Provides 28 analysis modules and telemetry-driven recommendations for optimal performance.
CLI & Python API: Accessible via command-line interface or Python API for flexible integration.

Quick Start

Run the obliteratus command with the model name to remove refusals:

obliteratus obliterate <model_name>

Frequently Asked Questions about obliteratus

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I remove refusal behaviors from an LLM without retraining or fine-tuning?▼

You can remove refusal behaviors from an LLM using mechanistic interpretability techniques like diff-in-means and SVD. This method modifies targeted model weights directly, bypassing the need for computationally expensive retraining or fine-tuning while preserving reasoning capabilities.

What is mechanistic interpretability for uncensoring language models?▼

Mechanistic interpretability for uncensoring language models involves analyzing internal activations to locate and ablate refusal directions. Techniques like diff-in-means and whitened SVD isolate specific refusal mechanisms, allowing targeted weight modification to bypass guardrails without degrading reasoning.

Do I need PyTorch and Transformers to ablate LLM refusal mechanisms?▼

Yes, ablating LLM refusal mechanisms requires PyTorch and Transformers, along with bitsandbytes, accelerate, and safetensors. These dependencies provide the necessary tensor operations, model loading, and weight serialization capabilities required for targeted model surgery.

What is the best way to remove guardrails from open-weight models while preserving reasoning?▼

The best way to remove guardrails while preserving reasoning is targeted weight ablation using diff-in-means and SVD. This mechanistic interpretability approach surgically removes refusal behaviors without retraining, maintaining the model's core expressive and reasoning capabilities.

Can I use a Python API to automate LLM uncensoring and model surgery?▼

Yes, you can automate LLM uncensoring and model surgery using a Python API or a command-line interface. Both access methods leverage the 28 analysis modules and mechanistic interpretability techniques to programmatically remove refusal behaviors from target models.