Communications of the ACM
ML System Based on Light Could Yield More Powerful, Efficient LLMs
With the new system, the team reports a greater than 100-fold improvement in energy efficiency and a 25-fold improvement in compute density over state-of-the-art digital computers for machine learning.
Credit: Ella Maru Studio
A team led by researchers at the Massachusetts Institute of Technology has developed a light-based machine learning system that could surpass the system behind ChatGPT in terms of power and efficiency, while also consuming less energy.
The compact architecture is based on arrays of vertical surface-emitting lasers developed by researchers at Germany's Technische Universitat Berlin.
The system uses hundreds of micron-scale lasers and the movement of light to perform computations.
The researchers said it could be scaled for commercial use in the near future, given its reliance on laser arrays commonly used in cellphone facial identification systems, and for data communication.
They found the system to be 100 times more energy efficient and 25 times more powerful in terms of compute density than current state-of-the-art supercomputers used to power existing machine learning models.
From MIT News
View Full Article
Abstracts Copyright © 2023 SmithBucklin, Washington, D.C., USA
No entries found