Instructions to use google/gemma-3-27b-it with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use google/gemma-3-27b-it with Transformers:

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

pipe = pipeline("image-text-to-text", model="google/gemma-3-27b-it")
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("google/gemma-3-27b-it")
model = AutoModelForMultimodalLM.from_pretrained("google/gemma-3-27b-it")
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]:]))

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use google/gemma-3-27b-it with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "google/gemma-3-27b-it"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "google/gemma-3-27b-it",
		"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/google/gemma-3-27b-it

SGLang

How to use google/gemma-3-27b-it 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 "google/gemma-3-27b-it" \
    --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": "google/gemma-3-27b-it",
		"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 "google/gemma-3-27b-it" \
        --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": "google/gemma-3-27b-it",
		"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 google/gemma-3-27b-it with Docker Model Runner:
```
docker model run hf.co/google/gemma-3-27b-it
```

Simple questions too hard?

#22

by urtuuuu - opened Mar 13, 2025

Discussion

urtuuuu

Mar 13, 2025

I wonder why this latest top model that has top chatbot arena score, fails at old simple questions like:

Making one candle requires 125 grams of wax and 1 wick. How many candles can I make with 500 grams of wax and 3 wicks? Be concise.
I have 50 books. I read 7 of them, how many books i have left?

That's just two examples from my test questions, but there are probably many more.

moekuro

Mar 14, 2025

Gemini 2.0 Flash can't answer this question either

urtuuuu

Mar 14, 2025

•

edited Mar 14, 2025

I mean, all this time i thought this question isn't hard at all. Almost all small models can answer it, probably even 3B or 4B models.

hoogenm

Mar 14, 2025

Q: Making one candle requires 125 grams of wax and 1 wick. How many candles can I make with 500 grams of wax and 3 wicks? Reason step by step. Then end the answer with a single digit.

A:
Here's how to solve this problem step-by-step:

Calculate how many candles can be made with the wax: 500 grams of wax / 125 grams/candle = 4 candles
Calculate how many candles can be made with the wicks: 3 wicks / 1 wick/candle = 3 candles
Determine the limiting factor: You can make 4 candles with the wax, but only 3 with the wicks. Therefore, the wicks are the limiting factor.

You can make 3 candles.

segaa

Mar 14, 2025

I've just tried the first question on Gemini 2.0 Flash and DeepSeek, both of them answered the question correctly (DeepSeek wasn't concise though).
The second question sounds ambiguous, this is, as an example, an answer form Perplexity.

Satya93

Mar 15, 2025

"Be concise." throws it off, if you don't add that it answers just fine. Example of bad prompting. Probably way more examples of bad prompts to add to this.

urtuuuu

Mar 15, 2025

•

edited Mar 15, 2025

"Be concise." throws it off, if you don't add that it answers just fine. Example of bad prompting. Probably way more examples of bad prompts to add to this.

It wasn't me who created this prompt, but all the models I tested in recent years usually answered correctly.
I just asked Phi-4, llama 3.1 8B, qwen 2.5 7B= all answered 3.
Second question about books, qwen turbo, 7b, 14b all answer 50 books left but 43 unread.
And of course all big models won't fail first question.

Renu11

Google org Apr 4, 2025

The "Gemma 3 27B model", as seen in below AI Studio screenshot proves its ability to accurately solve such problems.

urtuuuu

Apr 4, 2025

I know it's smart, but if the question isn't phrased "the right way", then...

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment