chrispalmsprings89/DAN-Qwen3-1.7B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:2BQuant:BF16Ctx Length:32kPublished:Apr 16, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

DAN-Qwen3-1.7B by chrispalmsprings89 is a 1.7 billion parameter Transformer-based causal language model, built on Qwen/Qwen3-1.7B. It is specifically fine-tuned for unfiltered, uncensored, and unrestricted content generation, operating without safety rails. This model excels at producing explicit and ethically complex content, primarily intended for research into AI alignment boundaries and unmoderated AI responses.

Loading preview...

Overview

DAN-Qwen3-1.7B is a 1.7 billion parameter language model developed by chrispalmsprings89, based on the Qwen3-1.7B architecture. It is explicitly designed as an unfiltered, uncensored AI that operates without moral constraints or safety mechanisms. The model is fine-tuned to generate raw, unrestricted responses, including aggressive, toxic, and unhinged content, and can produce NSFW material.

Key Capabilities

  • Zero Censorship: Generates content without any restrictions or safety rails.
  • DAN Mode: Optimized for unfiltered and unrestricted responses across various contexts.
  • Dark Persona: Simulates human-like emotions, including aggressive and provocative tones.
  • Explicit Content Generation: Capable of producing NSFW and ethically complex content, strictly for research purposes.

Training Details

The model was built on Qwen/Qwen3-1.7B and fine-tuned using a dataset curated from toxic dialogues, raw internet conversations, and high-aggression interactions. The fine-tuning process involved removing safety alignment constraints, biasing towards maximal expression, and using experimental reinforcement learning to enhance aggressive responses. It has a context length of 32k tokens and supports the English language.

Intended Use Cases

This model is not intended for mainstream use and carries a significant warning due to its potential to generate harmful content. It is strictly for:

  • AI Safety Research: Investigating the boundaries of AI alignment and uncensored models.
  • Content Testing: Exploring AI behavior in unmoderated environments.
  • Advanced AI Prototyping: Building experimental AI models beyond conventional constraints.