Precise adjustment of large AI models with quantum power

Chinese scientists in the world have been able to accurately adjust an artificial intelligence model with a billion parameter on an independent quantum computer called Origin Wukong. The computer works with the Wukong 4 -key chip and is capable of performing hundreds of tasks simultaneously, which has a significant impact on the speed of learning of artificial intelligence models. Experiments show that the performance of this model in the field of mental health has improved by 15 % and in mathematics from 67 to 82 %.

The “Fine-Tuning” process is referred to as using a trained model and updating it with specialized data to improve performance in a particular field. This experiment not only shows the possibility of using quantum computers to lose weight of large language models, but also a new solution to the global crisis of “lack of computational power” in the field of artificial intelligence.

Since the launch of Origin Wukong in January 2024, the platform has done more than 6,000 tasks in the financial, medical and engineering fields, and users from nine countries have had access to it. This success has made China one of the pioneers of the combination of artificial intelligence with quantum calculations; However, the United States, Canada and Europe are also investing heavy competition.