Ad
Short

Google's PaLM 2 gets better at math when you let the language model take a breather.

In a paper, researchers have investigated whether language models such as GPT-4 or PaLM 2 are suitable as optimizers that automatically find solutions to predefined problems, such as recommending movies or solving grade school math problems.

The language models tried to find the best prompts for each task on their own. The prompt that PaLM 2-L used particularly well to solve math problems was curious: "Take a deep breath and work on this problem step-by-step." Without taking a deep breath, accuracy dropped by almost ten points.

Ad
Ad
Ad
Ad
Short

Over 70 large language models with more than a billion parameters have been released in China, according to Baidu CEO Robin Li. Baidu joins other Chinese companies, including SenseTime and AI startups Baichuan Intelligent Technology, Zhipu AI, and MiniMax, that have launched AI chatbots after receiving regulatory approval. The latest version of Baidu's AI chatbot, Ernie 3.5, was recently released to the general Chinese public.

While over a billion parameters is not much, with some models breaking the trillion parameter barrier, there is still a place for smaller, specialized models, and it still shows the overall speed at which the Chinese AI industry is trying to bring commercially viable AI models to market. However, they are handicapped by hardware, as US regulations prevent the export of Nvidia's latest AI chips, the A100 and H100 to China and other countries. OpenAI's ChatGPT is not available in China.

Ad
Ad
Google News