Instructions to use tiny-random/gemma-3n with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tiny-random/gemma-3n with Transformers:

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

pipe = pipeline("image-text-to-text", model="tiny-random/gemma-3n")
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("tiny-random/gemma-3n")
model = AutoModelForImageTextToText.from_pretrained("tiny-random/gemma-3n")
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 tiny-random/gemma-3n with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tiny-random/gemma-3n"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tiny-random/gemma-3n",
		"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/tiny-random/gemma-3n

SGLang

How to use tiny-random/gemma-3n 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 "tiny-random/gemma-3n" \
    --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": "tiny-random/gemma-3n",
		"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 "tiny-random/gemma-3n" \
        --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": "tiny-random/gemma-3n",
		"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 tiny-random/gemma-3n with Docker Model Runner:
```
docker model run hf.co/tiny-random/gemma-3n
```

yujiepan commited on Jul 12, 2025

Commit

172dfa7

verified ·

1 Parent(s): 3e92e61

Upload folder using huggingface_hub

Browse files

Files changed (2) hide show

config.json +1 -5
preprocessor_config.json +1 -1

config.json CHANGED Viewed

@@ -10,7 +10,6 @@
     "conf_conv_kernel_size": 5,
     "conf_num_attention_heads": 2,
     "conf_num_hidden_layers": 2,
-    "conf_positional_bias_size": 256,
     "conf_reduction_factor": 4,
     "conf_residual_weight": 0.5,
     "gradient_clipping": 10000000000.0,
@@ -22,7 +21,6 @@
       128,
       32
     ],
-    "sscp_conv_eps": 0.001,
     "sscp_conv_group_norm_eps": 0.001,
     "sscp_conv_kernel_size": [
       [
@@ -71,7 +69,6 @@
     "altup_active_idx": 0,
     "altup_coef_clip": 120.0,
     "altup_correct_scale": true,
-    "altup_lr_multiplier": 1.0,
     "altup_num_inputs": 4,
     "attention_bias": false,
     "attention_dropout": 0.0,
@@ -100,7 +97,6 @@
     "num_hidden_layers": 4,
     "num_key_value_heads": 1,
     "num_kv_shared_layers": 2,
-    "query_pre_attn_scalar": 256,
     "rms_norm_eps": 1e-06,
     "rope_local_base_freq": 10000.0,
     "rope_scaling": null,
@@ -115,7 +111,7 @@
   "transformers_version": "4.54.0.dev0",
   "vision_config": {
     "architecture": "mobilenetv5_300m_enc",
-    "do_pooling": true,
     "hidden_size": 2048,
     "initializer_range": 0.02,
     "label_names": [

     "conf_conv_kernel_size": 5,
     "conf_num_attention_heads": 2,
     "conf_num_hidden_layers": 2,
     "conf_reduction_factor": 4,
     "conf_residual_weight": 0.5,
     "gradient_clipping": 10000000000.0,
       128,
       32
     ],
     "sscp_conv_group_norm_eps": 0.001,
     "sscp_conv_kernel_size": [
       [
     "altup_active_idx": 0,
     "altup_coef_clip": 120.0,
     "altup_correct_scale": true,
     "altup_num_inputs": 4,
     "attention_bias": false,
     "attention_dropout": 0.0,
     "num_hidden_layers": 4,
     "num_key_value_heads": 1,
     "num_kv_shared_layers": 2,
     "rms_norm_eps": 1e-06,
     "rope_local_base_freq": 10000.0,
     "rope_scaling": null,
   "transformers_version": "4.54.0.dev0",
   "vision_config": {
     "architecture": "mobilenetv5_300m_enc",
+    "do_pooling": false,
     "hidden_size": 2048,
     "initializer_range": 0.02,
     "label_names": [

preprocessor_config.json CHANGED Viewed

@@ -41,7 +41,7 @@
   "processor_class": "Gemma3nProcessor",
   "resample": 2,
   "rescale_factor": 0.00392156862745098,
-  "return_attention_mask": false,
   "return_tensors": null,
   "sampling_rate": 16000,
   "size": {

   "processor_class": "Gemma3nProcessor",
   "resample": 2,
   "rescale_factor": 0.00392156862745098,
+  "return_attention_mask": true,
   "return_tensors": null,
   "sampling_rate": 16000,
   "size": {