---
license: apache-2.0
pipeline_tag: image-text-to-text
base_model_relation: quantized
tags:
- heretic
- uncensored
- decensored
- abliterated
- gguf
base_model:
- SergiusFlavius/Qwen3-VL-4B-Instruct-heretic
---

# Qwen3-VL-4B-Instruct-heretic-gguf

## Overview

This repository provides a GGUF quantized builds of [Qwen3-VL-4B-Instruct-heretic](https://huggingface.co/SergiusFlavius/Qwen3-VL-4B-Instruct-heretic) for [llama.cpp](https://github.com/ggml-org/llama.cpp).

> This model is a decensored derivative of the official
> [Qwen/Qwen3-VL-4B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct),  
> modified using [Heretic](https://github.com/p-e-w/heretic) v1.1.0.

## Quantization Details

- **Backend:** `llama.cpp`
- **Commit:** `7537 (e68c19b0f)`
- **Method:** `Q4_K_M` for weights, `FP16` for multimodal adapter
- **Imatrix Optimization:** [x]️ *(custom dataset)*