Gemma is a family of open-weights Large Language Model (LLM) by Google DeepMind, based on Gemini research and technology.

This repository contains the implementation of the gemma PyPI package. A JAX library to use and fine-tune Gemma.

For examples and use cases, see our documentation. Please report issues and feedback in our GitHub.

1. Install JAX for CPU, GPU or TPU. Follow the instructions on the JAX website.

2. Run

   pip install gemma

Here is a minimal example to have a multi-turn, multi-modal conversation with Gemma:

from gemma import gm
# Model and parameters
model = gm.nn.Gemma3_4B()
params = gm.ckpts.load_params(gm.ckpts.CheckpointPath.GEMMA3_4B_IT)
# Example of multi-turn conversation
sampler = gm.text.ChatSampler(
    model=model,
    params=params,
    multi_turn=True,
)
prompt = """Which of the two images do you prefer?

Image 1: <start_of_image>
Image 2: <start_of_image>

Write your answer as a poem."""
out0 = sampler.chat(prompt, images=[image1, image2])
out1 = sampler.chat('What about the other image ?')

Our documentation contains various Colabs and tutorials, including:

• Sampling
• Multi-modal
• Fine-tuning
• LoRA
• ...

Additionally, our examples/ folder contain additional scripts to fine-tune and sample with Gemma.

• To use this library: Gemma documentation
• Technical reports for metrics and model capabilities:
  • Gemma 1
  • Gemma 2
  • Gemma 3
• Other Gemma implementations and doc on the Gemma ecosystem

To download the model weights. See our documentation.

Gemma can run on a CPU, GPU and TPU. For GPU, we recommend 8GB+ RAM on GPU for The 2B checkpoint and 24GB+ RAM on GPU are used for the 7B checkpoint.

We welcome contributions! Please read our Contributing Guidelines before submitting a pull request.

This is not an official Google product.