CoolFace
Modelpublic

Sinbad-The-Sailor/Qwen3.5-4B-NSFW-ARA-Heretic-Literotica

sourceHugging Faceupdated 6mo agoView on Hugging Face
9likes178downloads
Model Card

Qwen3.5-4B-NSFW-ARA-Heretic-Literotica

Overview

Qwen3.5-4B-NSFW-ARA-Heretic-Literotica is a specialized large language model built on the Qwen 3.5 architecture (4B parameters). It is designed for immersive erotic storytelling and creative prose, inheriting the technical uncensorship approaches of the ARA and Heretic frameworks.

The ARA-Heretic Lineage

This model is built upon the ARA (All-Roles-Allowed) and Heretic base. These frameworks are designed to minimize artificial refusals and maintain creative integrity during complex adult roleplay and storytelling. By building on this foundation, the model provides a more fluid and responsive experience for descriptive, NSFW content.

Training & Dataset

The model was fine-tuned on a curated corpus of 8,000 literary stories.

  • —Data Curation: The dataset was pre-filtered to prioritize high-quality prose and creative narratives. Many extreme tags and themes were intentionally excluded during the filtering process to ensure the model focuses on creative, story-driven erotica rather than shock-value content.
  • —Fine-tuning Method: Parameter-Efficient Fine-Tuning (LoRA)
  • —Target Modules: qproj, kproj, vproj, oproj, gateproj, upproj, down_proj (Comprehensive LoRA coverage)
  • —LoRA Hyperparameters: Rank (r): 8, Alpha: 16
  • —Optimization: Paged AdamW (32-bit) with a learning rate of 2e-5.

Usage & Inference

Prompt Format

This model uses the standard ChatML format: <|imstart|>system You are a creative and descriptive erotic storyteller.<|imend|> <|imstart|>user [Your prompt here]<|imend|> <|im_start|>assistant

Inference Settings

For optimal creative output, the following parameters are recommended:

  • —Temperature: 0.75 - 0.85
  • —Top-p: 0.90
  • —Repetition Penalty: 1.1

Responsible Use & Disclaimer

This model is intended for adult audiences (18+) and generates explicit NSFW content. It is built on uncensored foundations (ARA/Heretic) and is designed to follow creative prompts without artificial refusal. Users are responsible for ensuring their use of the model complies with local laws and regulations. The model is a tool for creative fiction and should not be used to generate illegal or non-consensual content.