Called Gemini, the highly anticipated AI model is capable of reasoning and understanding complex information with more nuance than Google's previous technology, according to Google. "This new model era represents one of the biggest scientific and engineering efforts we've undertaken as a company," Alphabet CEO Sundar Pichai said in a blog post.
Gemini includes three versions Gemini Ultra, Pro and Nano
Since OpenAI launched ChatGPT about a year ago, Google has been racing to produce AI software that rivals what the Microsoft-backed company has introduced. Google has added a new Gemini model to its Bard AI chatbot (specifically Gemini Nano) and said it plans to release the most advanced version of Gemini, Gemini Ultra, through Bard early next year. It also launched another version of Gemini called Gemini Pro.
Google says each version of Gemini is designed to use a different amount of processing power. The most powerful version is designed to run in data centers, and the smallest version will run efficiently on mobile devices, Google says.
DeepMind vice president Eli Collins said Gemini is the largest AI model his unit has ever created and is “significantly” cheaper to serve to users than the company’s previous larger machine learning models. As a result, it is not only more capable but also much more efficient. The latest model still requires a significant amount of computing power to train, he added, but Google is improving its process.
Alphabet also announced a new generation of its custom TPU AI chip, the Cloud TPU v5p, designed to train large AI models and clustered together in clusters of 8,960 chips. The new version of the customer processor can train large language models nearly three times faster than previous generations.
The company said the new chip is available to developers in “preview” form starting on December 6.
Source link
Comment (0)