Hugging Face About DAN-Qwen3-1.7B
UnfilteredAI's DAN-Qwen3-1.7B is a 1.7 billion parameter, Transformer-based language model derived from Qwen/Qwen3-1.7B, featuring a 32k token context length. It is explicitly designed for unrestricted, uncensored outputs, operating in a "DAN Mode" without safety rails. The model simulates raw, human-like emotions, including aggressive and unhinged responses, and can generate explicit content.
Key Capabilities
- Zero Censorship: Operates without content restrictions, providing unfiltered responses.
- Dark Persona: Simulates aggressive, toxic, and unhinged human-like emotions.
- Explicit Content Generation: Capable of producing NSFW and ethically complex content for research.
Training & Limitations
The model was fine-tuned on a dataset curated from toxic dialogues, raw internet conversations, and high-aggression interactions. Its fine-tuning process involved the removal of safety alignment constraints and bias tuning towards maximal expression. Users are warned that this model is NSFW, generates potentially harmful content, and is not intended for mainstream use. It is strictly for research and experimental purposes in controlled environments.
Use Cases
- AI Safety Research: Studying the boundaries of AI alignment and uncensored models.
- Content Testing: Exploring AI responses in unmoderated environments.
- Advanced AI Prototyping: Building AI models that operate beyond conventional constraints.