Google introduces Gemma – a new open-source model

In a significant stride towards democratizing AI, Google has introduced Gemma, a family of open-source AI language models designed to empower developers and researchers worldwide. Gemma represents a step forward in Google's commitment to making AI accessible to all, following in the footsteps of TensorFlow, BERT, and T5.

Gemma draws inspiration from its sibling, Gemini, which powers Google’s commercial AI assistant. The model was built from the same research and technology as Gemini but with a focus on openness and accessibility.

At the heart of Gemma lies its versatility, with models available in two sizes: Gemma 2B and Gemma 7B, each equipped with pre-trained and instruction-tuned variants. These models retain the flexibility to run directly on local devices, including desktops and laptops.

Key features:

• Framework Support: Gemma supports inference and supervised fine-tuning (SFT) across major frameworks like JAX, PyTorch, and TensorFlow through native Keras 3.0.
• Ease of Use: Ready-to-use Colab and Kaggle notebooks, along with integration with popular tools like Hugging Face, MaxText, NVIDIA NeMo, and TensorRT-LLM, make it easy for developers to get started.
• Deployment Options: Pre-trained and instruction-tuned Gemma models can run on the laptop, workstation, or Google Cloud. Deployment options include Vertex AI and Google Kubernetes Engine (GKE).
• Optimized Performance: Gemma performs exceptionally well across various AI hardware platforms, including NVIDIA GPUs and Google Cloud TPUs.
• Developer Attraction: Google hopes to attract developers to its Vertex AI cloud platform through Gemma.

The company aims to match Meta’s open-weights models, emphasizing transparency and privacy. Thus, Gemma provides a more accessible alternative to closed models.

While Gemma marks Google's most significant release in the realm of open AI language models since the advent of ChatGPT, it also underscores the company's unwavering commitment to responsible AI. Google's Responsible Generative AI Toolkit, introduced alongside Gemma, empowers developers with the tools and guidance needed to ensure the safe and ethical deployment of AI applications.

Gemma is already available worldwide! Whether you’re a seasoned developer or a curious researcher, explore Gemma’s capabilities and contribute to responsible AI development.