skills.rest
lxh755818-bot
lxh755818-botCOMMUNITY
Explore all skills by this author →
GitHub

obliteratus

Remove refusal behaviors from open-weight LLMs using diff-in-means and SVD.

Updated Apr 30, 2026
Search Tags:#LLM#mechanistic interpretability#refusal removal#model surgery#AI uncensoring
One-click install
npx skills add https://github.com/lxh755818-bot/obsidian-vault --skill obliteratus-lxh755818-bot
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: obliteratus
Source: https://github.com/lxh755818-bot/obsidian-vault/tree/main/backup/skills/mlops/inference/obliteratus
Command: npx skills add https://github.com/lxh755818-bot/obsidian-vault --skill obliteratus-lxh755818-bot

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?

This Skill addresses the issue of LLM refusal behaviors, allowing users to remove guardrails from models without retraining or fine-tuning.

Core Features & Use Cases

  • Refusal Removal: Eliminate refusal behaviors (guardrails) from open-weight LLMs.
  • Mechanistic Interpretability: Utilizes techniques like diff-in-means, SVD, and more to identify and remove refusal directions.
  • Use Case: When you want to create an uncensored version of an LLM without retraining, such as Llama, Qwen, or Mistral.

Quick Start

Run the obliteratus command with the obliterate subcommand followed by the model name to remove refusal behaviors.

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?▼

To remove refusal behaviors without retraining, you can use mechanistic interpretability techniques like diff-in-means and SVD to identify and surgically remove refusal directions from the model's weights.

Does obliteratus support uncensoring models like Llama, Qwen, and Mistral?▼

Yes, obliteratus supports uncensoring open-weight models like Llama, Qwen, and Mistral by identifying and eliminating their guardrails using mechanistic interpretability without requiring any fine-tuning.

What Python libraries do I need to perform LLM model surgery for AI uncensoring?▼

To perform LLM model surgery for AI uncensoring, you need Python libraries including torch, transformers, bitsandbytes, accelerate, and safetensors to execute the refusal removal analysis.

Can I eliminate LLM guardrails using SVD and diff-in-means techniques?▼

Yes, you can eliminate LLM guardrails using diff-in-means and SVD techniques, which allow you to precisely locate and remove the internal refusal directions within the model architecture.

What is the best way to create an uncensored LLM from an existing open-weight model?▼

The best way to create an uncensored LLM from an open-weight model is applying refusal removal via mechanistic interpretability, executing the obliterate subcommand to bypass retraining entirely.

Related Skills

Explore More Software Engineering Skills→
HuskyDannyHuskyDanny

obliteratus

Remove refusal behaviors from open-weight LLMs using mechanistic interpretability techniques.

Community
Advanced
aayushsoamaayushsoam
1

obliteratus

Remove refusal behaviors from LLMs using mechanistic interpretability techniques.

Community
Advanced
InverterNetworkInverterNetwork

obliteratus

Removes LLM refusal behaviors via mechanistic weight projection.

Official
Advanced
zeronx798zeronx798

obliteratus

Remove refusal behaviors from open-weight LLMs using SVD-based weight projection.

Community
Advanced