Instructions to use stamsam/FrankenGemma4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use stamsam/FrankenGemma4 with Transformers:

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

pipe = pipeline("text-generation", model="stamsam/FrankenGemma4")
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("stamsam/FrankenGemma4")
model = AutoModelForImageTextToText.from_pretrained("stamsam/FrankenGemma4")
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

vLLM

How to use stamsam/FrankenGemma4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "stamsam/FrankenGemma4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "stamsam/FrankenGemma4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/stamsam/FrankenGemma4

SGLang

How to use stamsam/FrankenGemma4 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 "stamsam/FrankenGemma4" \
    --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": "stamsam/FrankenGemma4",
		"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 "stamsam/FrankenGemma4" \
        --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": "stamsam/FrankenGemma4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use stamsam/FrankenGemma4 with Docker Model Runner:
```
docker model run hf.co/stamsam/FrankenGemma4
```

FrankenGemma4 Raw

FrankenGemma4 Raw is the source/archive repo for the FrankenGemma4 line.

This repo is intended to hold:

the raw unquantized checkpoint
provenance and lineage notes
benchmark references
optional raw source artifacts

stamsam/FrankenGemma

What This Repo Contains

The raw repo is meant for transparency and reproducibility, not as the easiest end-user download.

Lineage

The raw line comes from the same donor pair as the polished release:

arsovskidev/Gemma-4-E4B-Claude-4.6-Opus-Reasoning-Distilled
Jiunsong/supergemma4-e4b-abliterated

The polished MLX 4-bit release is the recommended default for most users, but this raw archive keeps the unquantized source checkpoint and provenance together for reproducibility.

Current Lead Branch

The current public lead branch in the project is Frankengemma4 V1.

Benchmark References

This archive repo keeps the benchmark evidence alongside the source checkpoint:

The detailed tables and summaries live in raw/provenance/benchmarks/.

Notes

Downloads last month: 200

Safetensors

Model size

1B params

Tensor type

BF16

U32

Model tree for stamsam/FrankenGemma4

Base model

google/gemma-4-E4B

Finetuned

google/gemma-4-E4B-it

Finetuned

Jiunsong/supergemma4-e4b-abliterated

Quantized

(14)

this model