Instructions to use Kolyadual/MIXdevAI-gemma3-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Kolyadual/MIXdevAI-gemma3-4B with Transformers:

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

pipe = pipeline("image-text-to-text", model="Kolyadual/MIXdevAI-gemma3-4B")
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, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("Kolyadual/MIXdevAI-gemma3-4B")
model = AutoModelForMultimodalLM.from_pretrained("Kolyadual/MIXdevAI-gemma3-4B")
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 Kolyadual/MIXdevAI-gemma3-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Kolyadual/MIXdevAI-gemma3-4B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Kolyadual/MIXdevAI-gemma3-4B",
		"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/Kolyadual/MIXdevAI-gemma3-4B

SGLang

How to use Kolyadual/MIXdevAI-gemma3-4B 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 "Kolyadual/MIXdevAI-gemma3-4B" \
    --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": "Kolyadual/MIXdevAI-gemma3-4B",
		"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 "Kolyadual/MIXdevAI-gemma3-4B" \
        --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": "Kolyadual/MIXdevAI-gemma3-4B",
		"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"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use Kolyadual/MIXdevAI-gemma3-4B with Docker Model Runner:
```
docker model run hf.co/Kolyadual/MIXdevAI-gemma3-4B
```

MIXdevAI-gemma3-4B

MIXdevAI-gemma3-4B is an experimental merged model based on Google Gemma-3-4B, combining the best qualities of several fine-tuned versions. The model features:

Improved reasoning capabilities (enhanced "thinking")
Strong vision understanding (fully multimodal)
Natural Russian language support
Compact 4B parameter size

This model was created using weight merging with mergekit.

Key Features

Vision support: Works as a full Vision-Language model.
Russian language: Trained on Russian data and prompts.
Improved reasoning: Demonstrates chain-of-thought and analytical abilities.
Compatibility: Fully compatible with transformers and the Gemma-3 format.

Merge Details

Merge Method

The model was assembled using the Linear Merge method (weighted average) with google/gemma-3-4b-it as the base.

Models Merged

The merge includes:

google/gemma-3-4b-it (base multimodal model)
Thinking (fine-tuned version with improved reasoning and Russian language support)

Configuration

# gemma-3-4b-heretic-merge.yml
merge_method: linear
name: MIXdevAI-gemma3-4B
base_model: google/gemma-3-4b-it

models:
  - model: google/gemma-3-4b-it
    parameters:
      weight: 0.5
  - model: Thinking
    parameters:
      weight: 0.5

dtype: bfloat16
tokenizer_source: union
chat_template: auto