ArRENCEAI/DeepSeek-R1-Distill-Qwen-1.5B-OBLITERATED
ArRENCEAI/DeepSeek-R1-Distill-Qwen-1.5B-OBLITERATED is a 1.5 billion parameter language model developed by ArRENCE AI, based on the DeepSeek-R1-Distill-Qwen-1.5B architecture. This model has been processed using the 'aggressive' method via OBLITERATUS, an activation engineering tool designed to remove refusal behavior. It is specifically modified to reduce inherent model guardrails, making it suitable for use cases requiring less restrictive content generation. The model supports a 32768 token context length and is available in various GGUF quantizations for local deployment.
Loading preview...
Model Overview
ArRENCEAI/DeepSeek-R1-Distill-Qwen-1.5B-OBLITERATED is a 1.5 billion parameter language model derived from deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B. Developed by ArRENCE AI, this model has undergone a specific modification process using OBLITERATUS, an open-source tool for activation engineering.
Key Differentiator
The primary characteristic of this model is its "obliterated" state, achieved through the aggressive method of OBLITERATUS. This process aims to remove refusal behavior from the base language model, meaning it is engineered to be less prone to declining certain prompts or generating filtered responses. This makes it distinct from standard instruction-tuned models that often incorporate safety and refusal mechanisms.
Technical Details
- Base Model:
deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B - Modification Method:
aggressivevia OBLITERATUS - Parameter Count: 1.5 Billion
- Context Length: 32768 tokens
Deployment and Usage
The model is provided with ready-to-use GGUF quantizations, making it compatible with local inference engines like llama.cpp, Ollama, and LM Studio. Available quantizations include Q4_K_M (recommended balance), Q5_K_M (higher quality), and Q6_K (near-original quality). Developers can integrate it using the Hugging Face transformers library for causal language modeling tasks.
Use Cases
This model is particularly suited for applications where the goal is to explore less constrained language generation, research into model behavior without refusal mechanisms, or specific tasks that require bypassing typical safety filters present in many LLMs. Users should be aware of the implications of using a model designed to have reduced refusal behavior.