---
base_model:
- OddTheGreat/Comet_12B_V.4
base_model_relation: quantized
pipeline_tag: image-text-to-text
tags:
- chat
- mlx
- apple
- 8bit
- multimodal
language:
- en
- ru
library_name: mlx
---
# Comet-12B V4 8-bit MLX (Uncensored Gemma-3 12B)

This is a merge of pre-trained language models:

- [`ReadyArt/The-Omega-Directive-Gemma3-12B-v1.0`](https://huggingface.co/ReadyArt/The-Omega-Directive-Gemma3-12B-v1.0)
- [`TheDrummer/Fallen-Gemma3-12B-v1`](https://huggingface.co/TheDrummer/Fallen-Gemma3-12B-v1)
- [`Delta-Vector/Pascal-12B`](https://huggingface.co/Delta-Vector/Pascal-12B)
- [`soob3123/amoral-gemma3-12B-v2`](https://huggingface.co/soob3123/amoral-gemma3-12B-v2)

The goal of this merge was to create a good, all‑purpose, **uncensored** model without excessive positive bias.


The model was converted to MLX format from [`OddTheGreat/Comet_12B_V.4`](https://huggingface.co/OddTheGreat/Comet_12B_V.4) using mlx-vlm version **0.1.23**.

Refer to the [original model card](https://huggingface.co/OddTheGreat/Comet_12B_V.4) for more details on the model.


## Use with mlx

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

```bash
python -m mlx_vlm.generate --model TheCluster/Comet-12B-v4-mlx-8bit --max-tokens 512 --temperature 0.0 --prompt "Describe this image." --image <path_to_image>
```