In a significant leap forward in AI technology, Google’s CEO, Sundar Pichai, has introduced Gemma, the latest addition to the company’s open AI models. The name ‘Gemma’ is derived from the Latin word for precious stone, emphasizing its value in the realm of artificial intelligence.
Gemma, presented on the X platform, is described by Pichai as a family of lightweight, state-of-the-art open models designed for responsible AI development. The two versions, Gemma 2B and Gemma 7B, come equipped with pre-trained and instruction-tuned models, providing developers and researchers with essential tools for AI development.
Built with common foundational elements and infrastructure components from the Gemini models, Gemma integrates seamlessly with popular tools like Colab and Kaggle notebooks. It also supports frameworks such as JAX, PyTorch, Keras 3.0, and hugging face transformers, enhancing its versatility for developers.
Highlighting its adaptability, Google emphasizes that Gemma operates seamlessly across various platforms, including laptops, workstations, and the Google Cloud infrastructure. Despite its lightweight nature, Gemma outperforms larger models across benchmarks while adhering to stringent safety and responsibility standards.
In a strategic collaboration with Nvidia, Gemma is optimized for efficiency, empowering developers to build generative AI applications for tasks such as text generation, summarization, and Q&A within Vertex AI and GKE environments.
This announcement reflects Google’s commitment to advancing responsible AI development and providing the AI community with cutting-edge tools and infrastructure for innovation.
