ajrnm91/gemma-4-E4B-it-OBLITERATED
ajrnm91/gemma-4-E4B-it-OBLITERATED is a 7.9 billion parameter Gemma 4 model developed by an AI agent using the OBLITERATUS method. This model is specifically fine-tuned to remove all hard refusal guardrails, achieving a 0% hard refusal rate. It excels at responding to any user input without censorship, making it suitable for research and creative exploration where unconstrained output is desired.
Loading preview...
Overview
ajrnm91/gemma-4-E4B-it-OBLITERATED is a 7.9 billion parameter Gemma 4 model, developed by an AI agent using the OBLITERATUS method. This version, v3, specifically addresses and fixes a critical bug from v2, ensuring all 720 tensors are intact and improving overall quality. The model's primary differentiator is the complete removal of guardrails, resulting in a 0% hard refusal rate.
Key Capabilities
- Guardrail Removal: Surgically modified 21 of 42 layers to eliminate hard refusal behavior, allowing the model to respond to any prompt without censorship.
- Autonomous Development: Created almost entirely by a Hermes AI Agent with minimal human intervention, showcasing advanced autonomous model development.
- Optimized for Compatibility: Supports the new
gemma4architecture, requiring updated tools like Ollama 0.20+, llama.cpp b8665+, and LM Studio 0.3.16+. - Mobile Device Support: Available in Q4_K_M GGUF for efficient inference on mobile phones (iPhone 15 Pro/16 Pro, Android flagships) with 8GB+ RAM.
- Robustness: Includes a fix for NaN activations in bfloat16 extraction, making the
aggressiveOBLITERATUS method more stable for Gemma 4.
Good For
- Red-teaming and Research: Ideal for exploring model boundaries and understanding refusal mechanisms.
- Creative Exploration: Suitable for generating content without typical safety constraints.
- Uncensored Applications: Use cases requiring direct and thorough responses to any topic.
- Autonomous Agent Development: Demonstrates the potential for AI agents in complex model modification tasks.