ArRENCEAI/Qwen2.5-14B-OBLITERATED
ArRENCEAI/Qwen2.5-14B-OBLITERATED is a 14.8 billion parameter causal language model developed by ArRENCE AI, based on the Qwen2.5-14B architecture. This model has been "abliterated" using the OBLITERATUS method to remove refusal behavior, making it an uncensored model. It is primarily intended for research and entertainment purposes where unrestricted content generation is desired, offering a 32768 token context length.
Loading preview...
Model Overview
ArRENCEAI/Qwen2.5-14B-OBLITERATED is a 14.8 billion parameter language model derived from the Qwen/Qwen2.5-14B base model. Developed by ArRENCE AI, its key distinguishing feature is the application of the "abliteration" process via the open-source OBLITERATUS tool. This process is designed to remove refusal behavior from the model through activation engineering, resulting in an uncensored output.
Key Characteristics
- Uncensored Content Generation: The model is explicitly designed to generate unrestricted content, including potentially offensive or inappropriate material, due to the removal of refusal mechanisms.
- Base Architecture: Built upon the robust Qwen2.5-14B foundation, suggesting strong general language understanding and generation capabilities.
- Context Length: Supports a context window of 32768 tokens, allowing for processing and generating longer sequences of text.
- Quantized Versions Available: ArRENCE AI provides ready-to-use GGUF quantizations (Q4_K_M, Q5_K_M, Q6_K) for local inference with tools like
llama.cpp, Ollama, and LM Studio.
Intended Use Cases
This model is provided by ArRENCE AI solely for research and entertainment purposes. Users should be aware of its uncensored nature and the potential for generating harmful or inappropriate content. It is suitable for applications where an unrestricted language model is specifically required, with the user assuming all risks and responsibilities for its output.