Instructions to use google/gemma-7b-it with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use google/gemma-7b-it with Transformers:

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

pipe = pipeline("text-generation", model="google/gemma-7b-it")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("google/gemma-7b-it")
model = AutoModelForCausalLM.from_pretrained("google/gemma-7b-it")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

llama-cpp-python

How to use google/gemma-7b-it with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="google/gemma-7b-it",
	filename="gemma-7b-it.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Inference
Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use google/gemma-7b-it with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf google/gemma-7b-it
# Run inference directly in the terminal:
llama-cli -hf google/gemma-7b-it

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf google/gemma-7b-it
# Run inference directly in the terminal:
llama-cli -hf google/gemma-7b-it

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf google/gemma-7b-it
# Run inference directly in the terminal:
./llama-cli -hf google/gemma-7b-it

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf google/gemma-7b-it
# Run inference directly in the terminal:
./build/bin/llama-cli -hf google/gemma-7b-it

Use Docker

docker model run hf.co/google/gemma-7b-it

LM Studio
Jan

vLLM

How to use google/gemma-7b-it with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "google/gemma-7b-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-7b-it",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/google/gemma-7b-it

SGLang

How to use google/gemma-7b-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-7b-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-7b-it",
		"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 "google/gemma-7b-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-7b-it",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use google/gemma-7b-it with Ollama:
```
ollama run hf.co/google/gemma-7b-it
```

Unsloth Studio new

How to use google/gemma-7b-it with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for google/gemma-7b-it to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for google/gemma-7b-it to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for google/gemma-7b-it to start chatting

Docker Model Runner
How to use google/gemma-7b-it with Docker Model Runner:
```
docker model run hf.co/google/gemma-7b-it
```

Lemonade

How to use google/gemma-7b-it with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull google/gemma-7b-it

Run and chat with the model

lemonade run user.gemma-7b-it-{{QUANT_TAG}}

List all available models

lemonade list

gemma-7b-it doesn't answer for some questions and returns '/n'

#55

by mudogruer - opened Feb 25, 2024

Discussion

mudogruer

Feb 25, 2024

•

edited Feb 25, 2024

The model doesn't generate any answer sometimes. I updated transformers library to '4.38.1' but nothing has changed.

EarthWorm001

Feb 26, 2024

Maybe we have the similar issue: https://huggingface.co/google/gemma-7b/discussions/33

suryabhupa

Google org Mar 4, 2024

Hey, Surya from the Gemma team here -- are you using the right formatting template and control tokens? What are your sampling settings?

mudogruer

Mar 4, 2024

Hey, thanks for respond. Here is the settings:
pipe = pipeline(
task = "text-generation",
model = fine_tuned_model,
tokenizer = tokenizer,
eos_token_id = model.config.eos_token_id,
max_new_tokens = 30,
)

def Sequence(promt):
sequence = pipe(
prompt,
max_new_tokens = 50,
temperature = 0.5
)
return sequence

prompt = ""+"Which is the heaviest element?"+""
print(Sequence(prompt))

suryabhupa

Google org Mar 4, 2024

This may not be using the right formatting such as <start_of_turn>, <end_of_turn> etc. -- does your prompt include those?

mudogruer

Mar 4, 2024

I guess it does.
def generate_prompt(sample):
full_prompt = f"""{sample['question']}
{sample['correct_answer']}
"""
return {"text": full_prompt}

It generates %40 of prompts. But mostly it doesn't. And it doesn't response exactly same prompts. I created a loop and created 200 prompts and run 3 times, got same results.

akshayparakh25

Mar 11, 2024

Hello @suryabhupa ,

I am trying to use gemma model for generation using langchain and hf text generation pipeline, and the output is far from expectation:

Code:

import transformers
from langchain_community.llms.huggingface_pipeline import HuggingFacePipeline

model_id = 'google/gemma-7b-it'
tokenizer = transformers.AutoTokenizer.from_pretrained(model_id, token=access_token,
    cache_dir='./../hf_models/')

model = transformers.AutoModelForCausalLM.from_pretrained(
    model_id,
    trust_remote_code=True,
    device_map='auto',
    token=access_token
)

generate_text = transformers.pipeline(
    model=model,
    tokenizer=tokenizer,
    # return_full_text=True,
    task="text-generation",
    generation_config=generation_config,
)

llm = HuggingFacePipeline(pipeline=generate_text)

prompt = '''<start_of_turn>user
Explain why the sky is blue<end_of_turn>
<start_of_turn>model'''
output = llm.invoke(prompt)
print(output)

Output:

The sky is blue because of the scattering of of particles particles particles.

When light particles scattering scattering scattering particles particles scattering scattering scattering scattering scattering particles particles scattering scattering scattering scattering scattering particles particles scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering scattering

Am I doing something incorrect? What should be the right way? Can you please share your comments.

suryabhupa

Google org Mar 15, 2024

Hey @akshayparakh25 thanks for raising this, that is certainly not what expected outputs should look like! @osanseviero this may be a problem on the HF side, does this look like the intended use case for this?

osanseviero

Google org Mar 17, 2024

Hi there! You're using a generation_config in your code example but are not sharing its value. This can significantly impact the generations, so I think the chosen values might be leading to this.

hrbrtrg

Mar 22, 2024

@akshayparakh25 When I use the base model I get the same weird response. When I try it with the it model, it gets solved. cc @osanseviero @suryabhupa

UserDAN

Apr 23, 2024

I'm having the same is you after fine-tuning the model with Qlora also I faced the same issue when I tried to use Gemma-7b give me the same behaviour with another weird behaviour which I mentioned here https://huggingface.co/google/gemma-7b/discussions/91

lkv

Google org Aug 28, 2025

•

edited Aug 28, 2025

Hi @mudogruer ,

The issue seems to stem from a combination of incorrect prompt formatting and sampling configuration. The Gemma-7B-IT model is instruction-tuned and expects structured prompts using <start_of_turn> and <end_of_turn> tokens. For consistent generation, ensure your prompt follows this format:

prompt = "<start_of_turn>user\nWhich is the heaviest element?\n<end_of_turn>\n<start_of_turn>model\n"

let us know if you have any concerns will assist you on this. Thank you.

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