Google just unveiled yet another family of AI models that can help streamline business and development for users. This family is called Gemma, and it consists of models formatted to fit on your computer.
It appears that Google is covering all of its bases. The company has the massive Gemini Pro and Gemini Ultra models that exist on gigantic servers. However, it also has Gemini Nano, a model that is small enough to fit on a smartphone. Well, Gemma is the middle ground between these extremes.
Gemma is the latest family of AI models from Google that can help you with AI development
If the name Gemma sounds familiar, it should. These models were built from the same technology used to develop Gemini. Gemma isn’t only one model, but it’s actually two. They come in two sizes: Gemma 2B and Gemma 7B. Both models are powerful, and they are both able to fit directly on your computer. So, they won’t be as powerful as the models like Gemini or OpenAI’s models. However, this still provides some incredibly powerful on-device AI for your specific needs.
Even though the Gemma models are designed to fit on your computer, Google states that they still outpace larger models in a few aspects. So, you’re still able to get some great performance.
Regardless of the size that you get, Gemma will still come with some great tools to help you with AI development. The company integrated Gemma with tools such as MaxText, Nvidia NeMo, and Hugging Face. These are popular and familiar tools that many developers are already familiar with. Along with that, the Gemma models are also optimized for Google Cloud, and it can be deployed in Vertex AI and Google Kubernetes Engine.
The company also assured us that these models also adhere to Google’s safety standards. So, you can rest assured that any outputs you get will be safe and acceptable. In fact, Gemma will come with a Responsible Generative AI Toolkit that will
2024-02-21 15:05:42