Pranavz/qwen-4b-2507-rp-mahou-nsfw

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:Apr 30, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

Pranavz/qwen-4b-2507-rp-mahou-nsfw is a 4 billion parameter Qwen3-based causal language model, fine-tuned by Pranavz for creative roleplay and character interaction. This model is a decensored version of its predecessor, specifically engineered to reduce refusals and enable more open-ended responses, making it suitable for applications requiring less restrictive content generation. It excels in vivid, action-asterisk style roleplay scenarios with a 32K context length.

Loading preview...

Overview

This model, Pranavz/qwen-4b-2507-rp-mahou-nsfw, is a 4 billion parameter language model based on the Qwen3 architecture. It is a decensored variant of Pranavz/qwen-4b-2507-rp-mahou, created using the Heretic v1.2.0 tool. The primary goal of this modification is to significantly reduce content refusals, as evidenced by its performance metric of 0/100 refusals compared to the original model's 99/100.

Key Capabilities

  • Decensored Output: Engineered to provide less restrictive content generation, making it suitable for use cases where the base model's safety filters might be too stringent.
  • Creative Roleplay: Fine-tuned using a full-sequence SFT method on the flammenai/flame-kindling-v1 dataset, which focuses on creative writing and character interaction.
  • Vivid Interaction: Optimized for generating vivid, descriptive text and handling actions denoted by asterisks, typical in roleplay scenarios.
  • Extended Context: Supports a context length of 32768 tokens, allowing for longer and more complex roleplay interactions.

Good For

  • Unfiltered Roleplay: Ideal for applications requiring a language model that will not refuse prompts based on content, offering greater creative freedom.
  • Character Interaction: Excels in generating dynamic and engaging dialogues and narratives for character-driven simulations.
  • Creative Writing: Suitable for generating imaginative stories, descriptions, and interactive fiction where a specific, vivid tone is desired.

Usage Notes

When using this model for roleplay, it is recommended to set enable_thinking=False in the chat template to prevent Chain-of-Thought (CoT) reasoning, which is generally not desired in direct roleplay. Recommended sampler settings include temperature between 0.7-0.85 and repetition_penalty between 1.05-1.15 to manage creativity and prevent loops.

Finetunes
1 models