---
license: gemma
base_model: wangzhang/gemma-4-26B-A4B-it-abliterix
language: en
pipeline_tag: image-text-to-text
library_name: mlx
tags:
  - mlx
  - abliterated
  - uncensored
  - gemma4
  - moe
  - direct-weight-editing
  - expert-granular-abliteration
---
# 🦆 zecanard/gemma-4-26B-A4B-it-uncensored-abliterix-MLX-4bit-nvfp4

[This model](https://huggingface.co/zecanard/gemma-4-26B-A4B-it-uncensored-abliterix-MLX-4bit-nvfp4) was converted to MLX from [`wangzhang/gemma-4-26B-A4B-it-abliterix`](https://huggingface.co/wangzhang/gemma-4-26B-A4B-it-abliterix) using `mlx-vlm` version **0.4.4**.
Please refer to the [original model card](https://huggingface.co/wangzhang/gemma-4-26B-A4B-it-abliterix) for more details.

## 🌟 Quality

Quantized vision language model with **4.843 bits per weight**.

`mlx_vlm.convert --quantize --q-bits 4 --q-group-size 16 --q-mode nvfp4`

## 🛠️ Customizations

This quant is aware of the current date, and also enables thinking (if available). You may disable this behavior by deleting the following line from the chat template:

`{%- set enable_thinking = true %}`

You may also need to adjust your environment’s **Reasoning Section Parsing** to recognize `<|channel>thought` as the **Start String**, and `<channel|>` as the **End String**.

## 🖥️ Use with `mlx`

```bash
pip install -U mlx-vlm
```

```bash
mlx_vlm.generate --model zecanard/gemma-4-26B-A4B-it-uncensored-abliterix-MLX-4bit-nvfp4 --max-tokens 100 --temperature 0 --prompt "Describe this image." --image <path_to_image>
```