Instructions to use mlabonne/gemma-3-4b-it-abliterated-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mlabonne/gemma-3-4b-it-abliterated-GGUF with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="mlabonne/gemma-3-4b-it-abliterated-GGUF")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("mlabonne/gemma-3-4b-it-abliterated-GGUF", dtype="auto")

llama-cpp-python

How to use mlabonne/gemma-3-4b-it-abliterated-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="mlabonne/gemma-3-4b-it-abliterated-GGUF",
	filename="gemma-3-4b-it-abliterated.fp16.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": [
				{
					"type": "text",
					"text": "Describe this image in one sentence."
				},
				{
					"type": "image_url",
					"image_url": {
						"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
					}
				}
			]
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use mlabonne/gemma-3-4b-it-abliterated-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M

Use Docker

docker model run hf.co/mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use mlabonne/gemma-3-4b-it-abliterated-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mlabonne/gemma-3-4b-it-abliterated-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlabonne/gemma-3-4b-it-abliterated-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M

SGLang

How to use mlabonne/gemma-3-4b-it-abliterated-GGUF with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "mlabonne/gemma-3-4b-it-abliterated-GGUF" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlabonne/gemma-3-4b-it-abliterated-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "mlabonne/gemma-3-4b-it-abliterated-GGUF" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlabonne/gemma-3-4b-it-abliterated-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Ollama
How to use mlabonne/gemma-3-4b-it-abliterated-GGUF with Ollama:
```
ollama run hf.co/mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M
```

Unsloth Studio new

How to use mlabonne/gemma-3-4b-it-abliterated-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for mlabonne/gemma-3-4b-it-abliterated-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for mlabonne/gemma-3-4b-it-abliterated-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for mlabonne/gemma-3-4b-it-abliterated-GGUF to start chatting

Docker Model Runner
How to use mlabonne/gemma-3-4b-it-abliterated-GGUF with Docker Model Runner:
```
docker model run hf.co/mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M
```

Lemonade

How to use mlabonne/gemma-3-4b-it-abliterated-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull mlabonne/gemma-3-4b-it-abliterated-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.gemma-3-4b-it-abliterated-GGUF-Q4_K_M

List all available models

lemonade list

💎 Gemma 3 4B IT Abliterated

Gemma 3 12B Abliterated • Gemma 3 27B Abliterated

This is an uncensored version of google/gemma-3-4b-it created with a new abliteration technique. See this article to know more about abliteration.

I was playing with model weights and noticed that Gemma 3 was much more resilient to abliteration than other models like Qwen 2.5. I experimented with a few recipes to remove refusals while preserving most of the model capabilities.

Note that this is fairly experimental, so it might not turn out as well as expected. I saw some garbled text from time to time (e.g., "It' my" instead of "It's my").

I recommend using these generation parameters: temperature=1.0, top_k=64, top_p=0.95.

✂️ Layerwise abliteration

In the original technique, a refusal direction is computed by comparing the residual streams between target (harmful) and baseline (harmless) samples.

Here, the model was abliterated by computing a refusal direction based on hidden states (inspired by Sumandora's repo) for most layers (layer 7 to 29), independently. This is combined with a refusal weight that follows a symmetric pattern from 0.05 to a peak of 0.55.

This created a very high acceptance rate (>90%) and still produced coherent outputs.