Google has announced the release of a new open AI based on the same research and technology that was used to build its Gemini AI models. The new open AI model is called Gemma and is aimed at the open community. The new model will help developers in building AI responsibly, as per the company’s claim. Google has also introduced the Responsible Generative AI Toolkit alongwith the new AI model, which contain a debugging tool and a guide with best practices for AI development based on the company’s experience.
The new AI model is offered in two different sizes: Gemma 2B and Gemma 7B. Both of these models come with pre-trained and instruction-tuned variants. These models are lightweight and can run directly on a developer’s laptop or desktop computer. In the term of key benchmarks, Gemma surpasses many larger models and they outperform other available open models, as per Google.
Apart from being powerful, these models are trained to be safe as the company has used automated techniques to remove personal information from the training data. In addition to this, the company has used human feedback in reinforcement learning to ensure models’ instruction-tuned variants show responsible behaviours. These models can be used by the companies and independent developers if they find any of the currently available AI models less powerful for their tasks.
Interested developers can access these models via data science platform Kaggle or Google Cloud. The company has planned to introduce more Gemma in the future for a diverse range of applications.
Meanwhile, Google has announced that some of its business centric AI tools will be based on its Gemini AI models. These models will be available at a lower price as the company is planning to compete with Microsoft-backed OpenAI.
Gemini for Google Workspace is the new name of Duet AI in Workspace. It is now priced at USD 20 or USD 30 per month, depending on the features selected by a single user.
ALSO READ: Gemini beats ChatGPT in multi-discipline tests by 3 per cent: Know-how
