Google has just launched Gemini, its long-awaited answer to ChatGPT | Trending Viral hub

[ad_1]

Google says there are three versions of the Gemini: Ultra, the largest and most capable; Nano, which is significantly smaller and more efficient; and Pro, medium size and medium capabilities.

Starting today, Google Bard, a chatbot similar to ChatGPT, will be powered by Gemini Pro, a change the company says will make it capable of more advanced reasoning and planning. Today, a specialized version of Gemini Pro is being incorporated into a new version of alpha code, a generative “research product” tool for coding from Google DeepMind. The most powerful version of Gemini, Ultra, will be included in Bard and available via a cloud API in 2024.

Sissy Hsiao, Google vice president and CEO of Bard, says the model’s multimodal capabilities have given Bard new abilities and made it better at tasks like summarizing content, brainstorming, writing, and planning. “These are the biggest quality improvements to Bard since its launch,” says Hsiao.

New Vision

Google showed several demos illustrating Gemini’s ability to handle problems related to visual information. The AI ​​model was seen responding to a video in which someone drew pictures, created simple puzzles, and asked for ideas for games involving a world map. Two Google researchers also showed how Gemini can help with scientific research by answering questions about a research paper that includes graphs and equations.

Collins says Gemini Pro, the model launching this week, outperformed the previous model that initially powered ChatGPT, called GPT-3.5, in six of eight benchmarks commonly used to test the intelligence of AI software.

Google says the Gemini Ultra, the model debuting next year, scores 90 percent, higher than any other model, including the GPT-4, in the Massive Multitasking Language Understanding (MMLU) benchmark, developed by academic researchers to test language models in questions on topics including mathematics, American history, and law.

“Gemini is the latest in a wide range of benchmarks: 30 of 32 of the most used in the machine learning research community,” Collins said. “And that’s why we see it establishing boundaries across the board.”

[ad_2]

Check Also

FIFA is said to be close to reaching a television deal with Apple for a new tournament | Trending Viral hub

[ad_1] FIFA, soccer’s world governing body, is close to a deal with Apple that would …

Tesla to recall Cybertruck in latest setback | Trending Viral hub

[ad_1] Tesla agreed to recall nearly 4,000 of its Cybertruck pickup trucks to repair an …

How scam calls and messages took over our daily lives | Trending Viral hub

[ad_1] Doctorow noted that just as the Internet has made routine tasks less onerous, it …

Leave a Reply

Your email address will not be published. Required fields are marked *