- Google announced Gemma, a new open model for artificial intelligence research.
- Gemma is a family of lightweight, cutting-edge open models designed for Google’s AI developers and researchers.
- Gemma is available in a variety of sizes and pre-trained variants, and responsible commercial use and distribution is permitted by Google.
Google takes a new step for artificial intelligence research Gemma He announced the open model he named. Gemma is a new “family of lightweight, cutting-edge open models” designed for Google’s AI developers and researchers. Gemma was developed by Google DeepMind and other internal teams and is inspired by the same research and technology used to create Gemini models. Its name means “precious stone” in Latin.
Gemma was available in several variants in sizes 2B and 7B. Google said it put Gemma through manual and automated testing, as well as “filtering certain personal information and other sensitive data from the training sets.” Gemma can run on laptops and desktops, as well as being compatible with IoT, mobile and cloud platforms. Google also noted that Gemma is compatible with tools like Vertex AI and Google Kubernetes Engine (GKE) and optimizes Google Cloud TPUs and Nvidia GPUs for easy deployment.
Google says the Gemma offers “best-in-class performance for its size” and “significantly outperforms larger models in key benchmarks.” Regarding the use of Gemma, Google stated that it allows “responsible commercial use and distribution for all organizations regardless of size.” Google also introduced a responsible generative AI toolkit that includes a debugging tool and security classification to examine Gemma’s performance and detect potential problems. While it is stated that Gemma is available worldwide, Google plans to expand its model family in the future.
Compiled by: Ayça Ayaz
