Qwen3.5-4B-NSFW-ARA-Heretic-Literotica-i1-GGUF

The i1 release introduces models quantized with an Importance Matrix, significantly improving performance on key prompt structures.

Overview

This repository contains GGUF (GPT-Generated Unified Format) versions of the Sinbad-The-Sailor/Qwen3.5-4B-NSFW-ARA-Heretic-Literotica model, a specialized 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.

These GGUF models have been quantized using an Importance Matrix (Imatrix), making them more robust and preserving key knowledge that is often lost in standard quantization.

What is GGUF?

GGUF is a binary format designed for single-file deployment of large language models, making it easy to use with tools like llama.cpp. It is a successor to the GGML format and offers better performance, flexibility, and metadata support.

The i1 Imatrix Quantization

The .i1. in the filenames signifies that these models were quantized using an Importance Matrix. This advanced technique measures the sensitivity of different weights in the neural network and uses this information to quantize the model more intelligently.

Benefits:

• Reduced Perplexity: Imatrix quantization significantly lowers perplexity compared to standard methods.
• Preserved Knowledge: It helps retain critical information and nuances that are often damaged during the quantization process.
• Higher Quality: The resulting models provide more coherent and contextually accurate outputs, especially for complex creative tasks.

Available Models

The following quantized models are available, offering a trade-off between performance and resource usage:

These files were renamed to include the .i1. tag to signify the Imatrix quantization. You may need to rename them locally if you have already downloaded the old files.

Usage & Inference

These models are designed to be used with llama.cpp and other GGUF-compatible backends.

Example with llama.cpp:

1. Download llama.cpp:

   git clone https://github.com/ggerganov/llama.cpp.git
   cd llama.cpp
   make
   

2. Download a GGUF model from this repository.
3. Run inference:

   ./main -m <path_to_gguf_model> -p "USER: [Your prompt here]
   

ASSISTANT:" -n 256 --temp 0.8 -c 4096 ```

Prompt Format

This model uses the standard ChatML format, adapted for a raw GGUF interface: <|im_start|>system You are a creative and descriptive erotic storyteller.<|im_end|> <|im_start|>user [Your prompt here]<|im_end|> <|im_start|>assistant

When using a text-based interface like llama.cpp, you can format your prompt like this:

<|im_start|>system
You are a creative and descriptive erotic storyteller.<|im_end|>
<|im_start|>user
She walked into the room, her eyes sparkling with a mischievous glint.<|im_end|>
<|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.