Instructions to use joerunde/gemma-4-moe-nano with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use joerunde/gemma-4-moe-nano with Transformers:

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

pipe = pipeline("image-text-to-text", model="joerunde/gemma-4-moe-nano")
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)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("joerunde/gemma-4-moe-nano")
model = AutoModelForMultimodalLM.from_pretrained("joerunde/gemma-4-moe-nano", device_map="auto")
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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use joerunde/gemma-4-moe-nano with vLLM:

Install from pip and serve model

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

SGLang

How to use joerunde/gemma-4-moe-nano 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 "joerunde/gemma-4-moe-nano" \
    --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": "joerunde/gemma-4-moe-nano",
		"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 "joerunde/gemma-4-moe-nano" \
        --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": "joerunde/gemma-4-moe-nano",
		"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 joerunde/gemma-4-moe-nano with Docker Model Runner:
```
docker model run hf.co/joerunde/gemma-4-moe-nano
```

gemma-4-moe-nano

A tiny, randomly-initialized stand-in for the gemma-4 MoE architecture (google/gemma-4-26B-A4B-it), built for per-architecture correctness testing of the spyre-inference backend on IBM Spyre accelerators. It is not a usable language model — the weights are random with only a brief conditioning pass, so its outputs are gibberish.

Why this exists

spyre-inference validates its Spyre lowering per model architecture. The real gemma-4 MoE checkpoints are 26B+ parameters and dominate CI wall-clock, yet no small public gemma-4 MoE checkpoint satisfies Spyre's head-size-multiple-of-64 rule (every public tiny-random one uses head_dim 16/32). This model fills that gap: it carries the exact gemma-4 architecture at a few-hundred-MB size, so the architecture can be exercised on every PR while the full product models run in an opt-in suite.

What is preserved from the real architecture

Gemma4ForConditionalGeneration / model_type: gemma4, text + vision tower
Dual local/global head dim (256 / 512, both multiples of 64) and the 2-KV-head global-attention layers
The 128-expert, top-8 MoE block
The sliding/full layer_types mix (with a full_attention layer retained)
The full 262144-token vocabulary and the real tokenizer

Shrunk (size only): hidden size, FFN / MoE intermediate sizes, text layer count (→ 6, keeping one full-attention layer), and vision layer count.

How it was generated

Reproducibly, by tests/data/generate_micro_gemma4_moe.py in spyre-inference: it loads the real config, overrides only the size fields, applies controlled init plus a short language-modeling conditioning pass (so the output distribution is non-flat and numerically well-behaved), and verifies a finite fp16 forward.

Validation

Loaded and run compiled on a Spyre card via the text-backbone path (hf_overrides={"architectures": ["Gemma4ForCausalLM"]}), with no CPU fallback — the shrunk config lowers through the same SpyreGemma4ForCausalLM per-block compile as the full model.

Downloads last month: 21

Safetensors

Model size

0.1B params

Tensor type

F16