Instructions to use lihyin/gemma3-siglip448 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use lihyin/gemma3-siglip448 with Transformers:

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

pipe = pipeline("image-text-to-text", model="lihyin/gemma3-siglip448")
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("lihyin/gemma3-siglip448")
model = AutoModelForMultimodalLM.from_pretrained("lihyin/gemma3-siglip448", 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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use lihyin/gemma3-siglip448 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "lihyin/gemma3-siglip448"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "lihyin/gemma3-siglip448",
		"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/lihyin/gemma3-siglip448

SGLang

How to use lihyin/gemma3-siglip448 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 "lihyin/gemma3-siglip448" \
    --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": "lihyin/gemma3-siglip448",
		"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 "lihyin/gemma3-siglip448" \
        --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": "lihyin/gemma3-siglip448",
		"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 lihyin/gemma3-siglip448 with Docker Model Runner:
```
docker model run hf.co/lihyin/gemma3-siglip448
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Model Card for Gemma3 4b with 448x448 SigLIP Image Encoder

Gemma3 4B with 448x448 SigLIP Vision Tower for edge deployment.

Model Details

This model is based on google/gemma-3-4b-it, whereas the 896x896 SigLIP in the original Gemma3 architecture is replaced with a 448x448 SigLIP model as it can be found in Efficient-Large-Model/paligemma-siglip-so400m-patch14-448. To fit the 256 image tokens, the average pooling layer at the output of the vision tower is adjusted for the 448x448 SigLIP model which has a smaller output embedding (1024 instead of 4096 patches in the 896x896 version). Details about the implementation can be found in the original Gemma3 Technical Report (Chapter 5.5. Vision encoder) [https://arxiv.org/pdf/2503.19786].

Downloads last month: -

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for lihyin/gemma3-siglip448

Base model

Efficient-Large-Model/paligemma-siglip-so400m-patch14-448

Finetuned

(8)

this model

Paper for lihyin/gemma3-siglip448

Gemma 3 Technical Report

Paper • 2503.19786 • Published Mar 25, 2025 • 59