“Brain-like” AI is said to run 100 times faster on ultra-long tasks using Chinese chips

The Chinese team has released what is called the world's first “brain-like” big language model – artificial intelligence A system designed to use less energy, improve performance, and operate without nvidia chip.

Developed by researchers at Beijing's Institute of Automation in China Academy of Sciences, Spikingbrain 1.0 mimics how the human brain fires only the neurons it needs.

So instead of activating the entire network like this chatgpt Other mainstream AI tools respond selectively to inputs, conserve power and accelerate response times.

Thanks to this design, this model can only learn less than 2% of the amount of traditional systems (less than 2% of traditional systems) while still being fast and efficient even when processing long texts.

In some cases, it ran 100 times faster than traditional models, according to a non-peer-reviewed technical paper not posted on the open access research repository Arxiv.

10:38

Hong Kong researchers develop AI models to detect fertile sperm with 96% accuracy

The system runs perfectly on China's homemade AI ecosystem powered by the Metax Chip platform, rather than Nvidia's dominant GPU hardware. This will make the model strategically important as the US will strengthen export controls for advanced AI chips.

Source link