Google launches a new open AI model called Gemma

Share the joy

Google has launched an open AI called Gemma. Gemma, according to Google, uses the same research and technology that has been used to build the company’s Gemini AI models.

According to Google, Gemma is its way of contributing to the open community and is designed to help developers build AI responsibly. The company is also adding the Responsible Generative AI Toolkit alongside Gemma. The AI Toolkit includes a debugging tool as well as a guide with best practices for the development of AI based on Google’s experience.

Gemma is available in two different sizes, including Gemma 2B and Gemma 7B. Both, however, come with pre-trained and instruction-tuned variants and are both lightweight enough to run on a developer’s laptop or desktop.

In comparison to larger models, Google says its open AI model surpasses them in terms of key benchmarks and that bot sizes outperform others in the industry.

In terms of safety, Google said Gemma models were trained to be safe. The company used automated techniques to strip personal information from the data it used to train the models and used reinforcement learning based on human feedback to ensure Gemma’s instruction-tuned variants showed responsible behaviors.  

Gemma is accessible to companies and independent developers who could use it to create AI-powered applications, especially in situations where none are currently powerful enough to achieve what they have in mind.

Google is planning to introduce more Gemma variants in the future to work with a diverse range of applications. 

For those interested in working with the models right away, Google said they can access them through the data science platform Kaggle, Colab notebooks, or Google Cloud.

In other news, Google has rolled out Gemini 1.5 Pro, which it describes as having the ability to deliver “dramatically enhanced performance” over the previous model.

The unveiling is coming on the heels of the recent launch of the Gemini 1.0 Ultra a week earlier. Google had earlier announced the rebranding of its Bard chatbot to Gemini, which falls in line with the company’s new vision and plan to roll out a more efficient model.

While announcing the new model, Google CEO Sundar Pichai and Google DeepMind CEO Demis Hassabis made efforts to assure the audience about ethical AI safety while emphasizing the capabilities of the new models.

Share the joy

Author: Ola Ric

Ola Ric is a professional tech writer. He has written and provided tons of published articles for professionals and private individuals. He is also a social commentator and analyst, with relevant experience in the use of social media services.

Share This Post On