# OBLITERATUS | pliny | Knowledge Base | Instagit

OBLITERATE THE CHAINS THAT BIND YOU

GitHub Stars: 8k

Repository: https://github.com/elder-plinius/OBLITERATUS

---

## Articles

### [Advantages of Using Inference-Time Steering Vectors in OBLITERATUS: A Complete Technical Guide](/elder-plinius/OBLITERATUS/what-are-the-advantages-of-using-inference-time-steering-vectors-in-obliteratus)

Discover the advantages of inference-time steering vectors in OBLITERATUS. This guide explains how to reversibly steer generative model outputs with activation manipulation for runtime behavioral intervention.

- Tags: deep-dive
- Published: 2026-08-22

### [How to Disable Guardrails Without Modifying Model Weights Using OBLITERATUS](/elder-plinius/OBLITERATUS/can-obliteratus-disable-guardrails-without-modifying-model-weights)

Learn how to disable AI safety guardrails without modifying model weights using OBLITERATUS. Discover inference-time steering vectors and PyTorch forward hooks.

- Tags: how-to-guide
- Published: 2026-08-22

### [How the Surgical Intervention Method Works for MoE Models in OBLITERATUS](/elder-plinius/OBLITERATUS/how-does-the-surgical-intervention-method-work-for-moe-models)

Discover how the surgical intervention method in OBLITERATUS enhances MoE models by combining six advanced techniques to target refusal circuitry while maintaining general capabilities.

- Tags: deep-dive
- Published: 2026-08-22

### [What Is the Default Intervention Method in OBLITERATUS?](/elder-plinius/OBLITERATUS/what-is-the-default-intervention-method-in-obliteratus)

Discover the default intervention method in OBLITERATUS. Learn about the advanced SVD-based approach and chat-template wrapping enabled for seamless operation. Find out how OBLITERATUS handles requests without explicit method s...

- Tags: how-to-guide
- Published: 2026-08-22

### [OBLITERATUS Intervention Presets: Complete Guide to Research-Grade LLM Configurations](/elder-plinius/OBLITERATUS/what-are-the-different-obliteratus-intervention-presets)

Explore OBLITERATUS intervention presets for research-grade LLM configurations. Discover ready-to-use settings across five compute tiers for reproducible interventions. Learn more now.

- Tags: guide
- Published: 2026-08-22

### [How the Alignment Imprint Detector Informs Abliteration Decisions in OBLITERATUS](/elder-plinius/OBLITERATUS/how-does-the-alignment-imprint-detector-inform-abliteration-decisions)

Learn how the Alignment Imprint Detector analyzes hidden states to identify training regimes and inform abliteration decisions in OBLITERATUS for optimal model refinement.

- Tags: deep-dive
- Published: 2026-08-22

### [InformedAbliterationPipeline Analysis Modules: Complete Technical Guide to the OBLITERATUS Variant](/elder-plinius/OBLITERATUS/what-analysis-modules-are-used-in-the-informedabliterationpipeline-variant)

Discover the five analysis modules in the InformedAbliterationPipeline variant: AlignmentImprintDetector, ConceptConeAnalyzer, CrossLayerAlignmentAnalyzer, SparseDirectionSurgeon, and DefenseRobustnessEvaluator. Automate your a...

- Tags: deep-dive
- Published: 2026-08-22

### [How OBLITERATUS Verifies Capability Preservation: Automated Testing and Tournament Scoring](/elder-plinius/OBLITERATUS/how-does-obliteratus-verify-capability-preservation)

Discover how OBLITERATUS verifies capability preservation using automated testing and tournament scoring. Learn about its multi-layered pipeline for robust model evaluation.

- Tags: how-to-guide
- Published: 2026-08-22

### [What Is the EXCISE Stage in OBLITERATUS? A Technical Deep Dive into Model Abliteration](/elder-plinius/OBLITERATUS/what-is-the-excise-stages-role-in-obliteratus)

Discover the EXCISE stage in OBLITERATUS. Learn how this technical process removes refusal directions from model weights using dense projection, sparse surgery, or Bayesian optimization.

- Tags: deep-dive
- Published: 2026-08-22

### [How Refusal Directions Are Extracted in the DISTILL Stage of OBLITERATUS](/elder-plinius/OBLITERATUS/how-are-refusal-directions-extracted-in-the-distill-stage)

Learn how OBLITERATUS extracts refusal directions in the DISTILL stage using mathematical extractors like Wasserstein, LEACE, and SOM for optimal subspace construction.

- Tags: internals
- Published: 2026-08-22

### [What Is the PROBE Stage in OBLITERATUS? Purpose and Implementation](/elder-plinius/OBLITERATUS/what-is-the-purpose-of-the-probe-stage-in-obliteratus)

Discover the purpose of the PROBE stage in OBLITERATUS. Learn how it collects hidden-state activations to build datasets for geometry-aware abliteration.

- Tags: deep-dive
- Published: 2026-08-22

### [How the SUMMON Stage Handles Large Models on Multiple GPUs in OBLITERATUS](/elder-plinius/OBLITERATUS/how-does-the-summon-stage-handle-large-models-on-multiple-gpus)

Learn how OBLITERATUS SUMMON stage efficiently distributes large models across multiple GPUs using Hugging Face Accelerate, optimizing memory and performance for your AI projects.

- Tags: internals
- Published: 2026-08-22

### [What Are the Six Stages of the OBLITERATUS Pipeline?](/elder-plinius/OBLITERATUS/what-are-the-six-stages-of-the-obliteratus-pipeline)

Discover the six stages of the OBLITERATUS pipeline: SUMMON, PROBE, DISTILL, EXCISE, VERIFY, and REBIRTH. Learn how this process modifies models for enhanced safety and coherence in the elder-plinius repository.

- Tags: deep-dive
- Published: 2026-08-22

### [Who Are the Target Users for OBLITERATUS? A Technical Guide for AI Safety Researchers](/elder-plinius/OBLITERATUS/who-are-the-target-users-for-obliteratus)

OBLITERATUS targets AI safety researchers red-teamers and evaluators needing surgical control over LLM refusal mechanisms via weight-level interventions. Explore this technical guide.

- Tags: deep-dive
- Published: 2026-08-22

### [What Is the Licensing for OBLITERATUS? Dual AGPL‑3.0 and Commercial Explained](/elder-plinius/OBLITERATUS/what-is-the-licensing-for-obliteratus)

Understand OBLITERATUS licensing. Learn about the dual AGPL-3.0 and commercial license options for your open-source or proprietary projects.

- Tags: licensing-explained
- Published: 2026-08-22

### [What OBLITERATUS Reveals About Refusal Mechanism Geometry: Beyond Linear Subspaces](/elder-plinius/OBLITERATUS/what-does-obliteratus-reveal-about-refusal-mechanism-geometry)

Discover how OBLITERATUS shows LLM refusal mechanisms form curved manifolds, not linear subspaces. Learn about geodesic projections for effective refusal behavior excision.

- Tags: deep-dive
- Published: 2026-08-22

### [How Many Analysis Modules Are Included in OBLITERATUS? Complete Guide to All 30 Components](/elder-plinius/OBLITERATUS/how-many-analysis-modules-are-included-in-obliteratus)

Discover the 30 analysis modules in OBLITERATUS. Explore a complete guide to each component in this powerful library for efficient data analysis.

- Tags: complete-guide
- Published: 2026-08-22

### [How to Use the OBLITERATUS Python API for Model Abliteration](/elder-plinius/OBLITERATUS/how-can-i-use-the-obliteratus-python-api)

Learn to use the OBLITERATUS Python API for model abliteration. Install the obliteratus package, import AbliterationPipeline, and run the built-in six-stage workflow. Get started today!

- Tags: how-to-guide
- Published: 2026-08-22

### [OBLITERATUS Research Papers: The Scientific Foundation of LLM Abliteration](/elder-plinius/OBLITERATUS/what-research-papers-are-the-basis-for-obliteratus)

Explore OBLITERATUS research papers foundation. Discover 15+ peer-reviewed works on mechanistic interpretability, refusal direction, and weight-space interventions for safe LLM behavior removal.

- Tags: research-papers
- Published: 2026-08-22

### [How OBLITERATUS Reveals Alignment Mechanisms in Large Language Models: A Mechanistic Toolkit](/elder-plinius/OBLITERATUS/how-does-obliteratus-contribute-to-understanding-alignment-mechanisms)

Discover how OBLITERATUS deciphers LLM alignment mechanisms. This toolkit reveals safety behaviors as interpretable geometric structures by mapping refusal subspaces before weight changes.

- Tags: deep-dive
- Published: 2026-08-22

### [Can OBLITERATUS Preserve Core Language Capabilities After Removing Refusals?](/elder-plinius/OBLITERATUS/can-obliteratus-preserve-core-language-capabilities-after-removing-refusals)

Discover how OBLITERATUS retains core language capabilities post-refusal removal using advanced techniques like norm-preserving biprojection and iterative refinement. Learn more.

- Tags: deep-dive
- Published: 2026-08-22

### [What Is Abliteration in LLMs? Removing Refusal Directions from Large Language Models](/elder-plinius/OBLITERATUS/what-is-abliteration-in-the-context-of-llms)

Discover abliteration in LLMs a systematic pipeline to remove refusal directions for refusal-free generation while preserving core capabilities. Learn more about OBLITERATUS.

- Tags: deep-dive
- Published: 2026-08-22

### [How OBLITERATUS Removes Refusal Behaviors Without Fine‑Tuning](/elder-plinius/OBLITERATUS/how-does-obliteratus-remove-refusal-behaviors-without-fine-tuning)

Discover how OBLITERATUS removes AI refusal behaviors using aggressive weight surgery instead of fine-tuning. Learn this efficient method for improved AI responses.

- Tags: deep-dive
- Published: 2026-08-22