Google launches open source Gemma AI model

Google has released a new open-source large language model (LLM) called Gemma to help the open community of developers and researchers powering artificial intelligence (AI) innovation.

Built on the technology behind Google’s powerful Gemini models, the Gemma models aim to bring sophisticated AI tools to everyone, as they are capable of running on laptops and other devices with limited computing power.

It has been developed by Google DeepMind and other teams across Google.

Available in two sizes, Gemma 2B and Gemma 7B, the models aim to provide users with the ability to develop applications such as chatbots, content generators, and other language-related tasks. 

This accessibility could potentially attract developers and organizations who previously lacked access to such resources.

The tech giant is also prioritizing responsible use with the Responsible Generative AI Toolkit. This suite of tools acts as a safety net for developers and researchers, ensuring their AI applications built with Gemma are ethical and sound.

A novel methodology allows them to create robust safety filters, even with limited data, ensuring Gemma’s outputs stay within acceptable boundaries and don’t cause harm. 

The toolkit also helps identify issues before they arise. A dedicated debugging tool dives into Gemma’s inner workings, allowing developers to pinpoint and address any potential problems before they cause harm.

Finally, the toolkit also equips developers with the knowledge of seasoned AI experts. Google, with its vast experience in large language models, will share best practices directly within the toolkit. 

The company shared that Gemma is a significant step forward in the development of open-source language models. With its lightweight design and powerful capabilities, it has the potential to be used in a wide range of applications, from chatbots and virtual assistants to machine translation and text generation.

Google CEO Sundar Pichai took to X (formerly Twitter) to unveil the Gemma models. 

“Introducing Gemma – a family of lightweight, state-of-the-art open models for their class built from the same research & tech used to create the Gemini models. 

Demonstrating strong performance across benchmarks for language understanding and reasoning, Gemma is available worldwide starting today in two sizes (2B and 7B), supports a wide range of tools and systems, and runs on a developer laptop, workstation or  @GoogleCloud,” said Pichai. 

Introducing Gemma – a family of lightweight, state-of-the-art open models for their class built from the same research & tech used to create the Gemini models.

Demonstrating strong performance across benchmarks for language understanding and reasoning, Gemma is available… pic.twitter.com/qWDAmEw7CP

— Sundar Pichai (@sundarpichai) February 21, 2024

“Accompanying our model weights, we’re also releasing tools to support developer innovation, foster collaboration, and guide responsible use of Gemma models,” Google said.