Google released an open AI model Gemmasaid it was created using the same research and technology used to build it Gemini AI models. The company says Gemma is its contribution to the open community and is designed to help developers “build AI responsibly.” So, he and Gemma also launched the Responsible Generative AI Toolkit. It includes a debugging tool as well as a guide to best practices for AI development based on Google’s expertise.
The company offers Gemma in two different sizes—the Gemma 2B and the Gemma 7B—both with pre-trained and manual-tuned options, and both are light enough to run directly on a developer’s laptop or desktop. Google says that the Gemma outperforms larger models when it comes to key benchmarks, and both models outperform other outdoor models by size.
As well as being strong, Gemma models are trained to be safe. Google used automated techniques to remove personal information from the data it used to train the models, and applied reinforcement learning with human feedback to ensure that Gemma’s instructional variants behaved responsibly. Companies and independent developers can use Gemma to build AI-powered apps, especially if none of the existing open models are powerful enough for what they want to build.
Google plans to introduce more Gemma variants for more different apps in the future. However, those who want to start working with the models right now can access them through data science platform Kaggle, the company’s Colab laptops, or Google Cloud.
