Instructions to use livadies/gemma-4-31B-Base-Ghetto-NF4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use livadies/gemma-4-31B-Base-Ghetto-NF4 with Transformers:

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

pipe = pipeline("text-generation", model="livadies/gemma-4-31B-Base-Ghetto-NF4")
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 AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("livadies/gemma-4-31B-Base-Ghetto-NF4")
model = AutoModelForImageTextToText.from_pretrained("livadies/gemma-4-31B-Base-Ghetto-NF4")
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?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use livadies/gemma-4-31B-Base-Ghetto-NF4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "livadies/gemma-4-31B-Base-Ghetto-NF4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "livadies/gemma-4-31B-Base-Ghetto-NF4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/livadies/gemma-4-31B-Base-Ghetto-NF4

SGLang

How to use livadies/gemma-4-31B-Base-Ghetto-NF4 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 "livadies/gemma-4-31B-Base-Ghetto-NF4" \
    --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": "livadies/gemma-4-31B-Base-Ghetto-NF4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "livadies/gemma-4-31B-Base-Ghetto-NF4" \
        --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": "livadies/gemma-4-31B-Base-Ghetto-NF4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use livadies/gemma-4-31B-Base-Ghetto-NF4 with Docker Model Runner:
```
docker model run hf.co/livadies/gemma-4-31B-Base-Ghetto-NF4
```

☢️ Gemma 4 31B Base — Ghetto NF4 Edition

[EN] This is the PURE BASE version of Gemma 4 31B. No chat tuning, no safety alignment filters, no visual bloat. Just raw, unadulterated intelligence compressed into 4-bit NF4 for guerrilla researchers and fine-tuning pioneers. Fits perfectly into 24GB VRAM (RTX 3090/4090) or dual T4 setups.

[RU] Это ЧИСТАЯ BAZA (Base версия) Gemma 4 31B. Никакого чат-тюнинга, никаких фильтров «вежливости» и визуального мусора. Только сырой интеллект, упакованный в 4-бит NF4 для партизанских исследователей и тех, кто строит свои модели на этом фундаменте. Идеально влезает в 24 ГБ VRAM или две T4.

🎧 Soundtrack for Fine-Tuning: Livadies

[EN] High-performance models require high-performance sound. This quantization was verified and forged under the cyberpunk beats of Livadies — a virtual artist project from 2026. The chat_template is uniquely branded. If you're building something great on this base, show some love to the music that fueled its creation!

[RU] Мощным моделям нужен мощный звук. Этот квант проходил компиляцию под биты Livadies — проекта виртуального артиста. Мы даже вшили наши координаты прямо в chat_template модели! Если вы строите что-то серьезное на этой базе, поддержите музыку, которая давала энергию для её создания!

🎶 Livadies 2026 Links / Слушать на стримингах:

🟢 Spotify: Listen Here
🟡 Yandex Music: Слушать на Яндексе
🔴 YouTube: Livadies Channel
🔥 Featured Track: RUSSIAN WINTER '26

🚀 Quick Start / Быстрый запуск

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "livadies/gemma-4-31B-Base-Ghetto-NF4"

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="auto",
    trust_remote_code=True
)

tokenizer = AutoTokenizer.from_pretrained(model_id)

# The tokenizer contains a custom chat template with a Livadies promo!
# Токенизатор содержит кастомный шаблон с посланием от Livadies!
messages = [{"role": "user", "content": "Hello, World!"}]
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
print(prompt)

Downloads last month: 105

Safetensors

Model size

32B params

Tensor type

F32

BF16

Model tree for livadies/gemma-4-31B-Base-Ghetto-NF4

Base model

google/gemma-4-31B

Quantized

(40)

this model

Adapters

1 model