[ad_1]
Google has released Gemma 2B and 7BA pair of open-source AI models that let developers use that research Its main gemini more freely. While Gemini is a larger closed AI model that competes directly with OpenAI’s ChatGPT (and is almost as powerful), the lighter Gemma will likely be better suited for smaller tasks like simple chatbots or summaries.
But what these models lack in complexity they make up for in speed and cost of use. Despite its small size, Google claims that Gemma models “outperform significantly larger models on key benchmarks” and “are able to run directly on developer laptops or desktop computers.” They will be available through Kaggle, Hugging Face, Nvidia’s Nemo, and Google’s Vertex AI.
The release of Gemma in the open-source ecosystem is completely different from the way Gemini is released. While developers can build on Gemini, they do so either through the API or by working on Google’s Vertex AI platform. Gemini is considered a closed AI model. By making Gemma open source, more people can experiment with Google’s AI rather than turning to competitors that offer better access.
Both model sizes will be available with commercial licenses regardless of the size of the organization, number of users, and type of project. However, Google – like other companies – Often prevents its models from being used for specific tasks such as weapons development programs.
Gemma will also ship with a “responsible AI toolkit”, as guardrails can be harder to implement in an open model than in a more closed system like Gemini. Tris Warkentin, director of product management at Google DeepMind, said the company “has done more extensive red-teaming for Gemma because of the inherent risks associated with open models.”
The Responsible AI Toolkit will allow developers to create their own guidelines or restricted word lists when deploying Gemma in their projects. It also includes a model debugging tool that lets users examine Gemma’s behavior and fix problems.
According to Warkentin, the models currently work best for linguistic tasks in English. “We hope we can work together with the community to meet the needs of the market in addition to English-language works,” he told reporters.
Developers can use Gemma at Kaggle for free, and first-time Google Cloud users get a $300 credit for using the model. The company said researchers can apply for up to $500,000 in cloud credits.
While it’s unclear how much demand there is for smaller models like Gemma, other AI companies have also released lighter-weight versions of their flagship Foundation models. meta kicked out llama 2 7b, the smallest iteration of last year’s Llama 2. Gemini itself comes in several sizes, including Gemini Nano, Gemini Pro, and Gemini Ultra, and Google recently announced a fiery gemini 1.5 – Again, for business users and developers for now.
By the way, Gemma means precious stone.
[ad_2]
Thanks For Reading