Chinese scientists have used the Origin Wukong quantum computer to fine-tune a large language model with a billion different parameters.
It is noted that Origin Wukong is powered by a 72-qubit superconducting quantum chip. This quantum chip allows one data set to perform hundreds of different tasks simultaneously. This, in particular, significantly speeds up training of AI language models.
This chip has been successfully tested on a mental health chatbot dataset. It is emphasized that learning process was accelerated by 15%. The model’s accuracy increased from 68% to 82%. While the model itself was reduced by 76%, its performance increased by 8.4%.
According to analysts, the experiment not only demonstrated the possibility of using quantum computing to reduce large language models, but also suggested a way to solve the problem of lack of computing power for AI training. Since its launch in January 2024, the Origin Wukong quantum computer has already completed more than 350 thousand tasks, from medical diagnostics to financial analytics and engineering design. Experts from 139 countries have used the services of this computer.
Fine-tuning is the process of taking a pre-trained language model and adapting it to perform better on specific tasks or in a specific industry. Instead of training a model from scratch, fine-tuning uses existing knowledge as a foundation and improves it with specialized data, such as legal documents, medical records, or financial reports.
This greatly improves the model’s accuracy, relevance, and effectiveness when applied to real-world conditions. By adapting the model to understand industry language, terminology, and context, companies can unlock more reliable and accurate AI-based solutions while reducing development time and cost.
The next step is to technology application to even more complex artificial intelligence systems, including those that work simultaneously with text, images, and sound.
According to researchers close to the latest test, Origin Wukong is one of the first quantum platforms to handle a full-scale AI fine-tuning workload with tangible results.
Meta has launched the first two Llama 4 models: multimodal, fast and powerful — two more on the way
Source: Interesting Engineering
Spelling error report
The following text will be sent to our editors: