Responsible by design
Gemma is designed with our AI Principles at the forefront. To ensure the safety and reliability of Gemma’s pre-trained models, we employed automated techniques to filter out sensitive data from training sets. Additionally, we utilized fine-tuning and reinforcement learning from human feedback (RLHF) to align our models with responsible behaviors. Our evaluations, including manual red-teaming, automated adversarial testing, and assessments of model capabilities for risky activities, are detailed in our Model Card.
We are also launching a new Responsible Generative AI Toolkit alongside Gemma to assist developers and researchers in prioritizing the development of safe and responsible AI applications. The toolkit includes:
- Safety classification: A methodology for building robust safety classifiers with minimal examples.
- Debugging: A tool for investigating Gemma’s behavior and addressing potential issues.
- Guidance: Best practices for model builders based on Google’s experience in developing large language models.
Optimized across frameworks, tools and hardware
Gemma models can be fine-tuned on your own data to meet specific application requirements, such as summarization or retrieval-augmented generation (RAG). Gemma is compatible with a variety of tools and systems:
- Multi-framework tools: Support for inference and fine-tuning across Keras 3.0, PyTorch, JAX, and Hugging Face Transformers.
- Cross-device compatibility: Gemma models run on various device types, including laptops, desktops, IoT devices, mobile devices, and the cloud.
- Cutting-edge hardware platforms: Gemma is optimized for NVIDIA GPUs, offering high performance and integration with the latest technology.
Free credits for research and development
Gemma is designed for developers and researchers in the AI community. You can start using Gemma today with free access in Kaggle, a free tier for Colab notebooks, and $300 in credits for first-time Google Cloud users. Researchers can also apply for Google Cloud credits up to $500,000 to accelerate their projects.
Getting started
For more information on Gemma and quickstart guides, visit ai.google.dev/gemma.
Stay tuned for new variants of the Gemma model family for various applications. Join us in upcoming events and opportunities to connect, learn, and build with Gemma.
We look forward to seeing your creations!
