Google has been playing a catching game ever since OpenAI released ChatGPT, a large language processing tool built on AI technology, almost a year ago. Soon they launched Bard, but the public reception was rather lukewarm. Google has now revealed Gemini, which they claim to be the most powerful AI model the world has ever seen. It comes in three sizes -ultra, pro, and nano.
“One of the powerful things about this moment,” Pichai said “is you can work on one underlying technology and make it better and it immediately flows across our products.” Here’s everything you need to know about Gemini, the much-awaited answer from Google to GPT4.
What Is Gemini?
Gemini is the latest in the line of Large Language Models (LLM) technologies, from Google’s AI division DeepMind. Google describes it as “natively multimodal”, meaning it is trained not only in text format but also in code, images and audio. Truly, a jack of all trades but this feature is available only in Gemini Ultra, the premium model that will be available next year. OpenAI also has models to read images and audio but they are separate. With Gemini, Google is trying to introduce a multi-sensory model.
Gemini Is Powered By TPU
Gemini is powered by homegrown AI accelerator technology called tensor processing units (TPUs). TPUs are more efficient and significantly faster than their previous iteration like PaLM. Google also confirmed the launch of Cloud TPU v5p.
“In our early stage usage, Google DeepMind and Google Research have observed 2X speedups for LLM training workloads using TPU v5p chips compared to the performance on our TPU v4 generation,” says Jeff Dean, chief scientist, Google DeepMind and Google Research. “The robust support for ML Frameworks (JAX, PyTorch, TensorFlow) and orchestration tools enables us to scale even more efficiently on v5p. With the 2nd generation of SparseCores, we also see significant improvement in the performance of embeddings-heavy workloads.”
Know About All Three Versions of Gemini
Gemini Nano: This version is specifically for Android-powered smartphones. The Nano version lets you summarise the research papers, suggest texts while chatting, and can also transcribe the recordings.
Gemini Pro: Gemini Pro will be integrated into Google’s AI chartbot, Bard, making it a serious rival to ChatGPT.
Gemini Ultra: This is the most advanced version which will be available next year, once the company is satisfied with the testing process.
How To Access Gemini
The Pixel 8 phone is powered by Gemini, and most of the offerings from Google, such as their web browser, and the search console, will be integrated with Gemini Pro or Nano version.
