no code implementations • 9 Nov 2021 • Daniel Nichols, Siddharth Singh, Shu-Huai Lin, Abhinav Bhatele
This phenomenon has spurred the development of algorithms for distributed training of neural networks over a larger number of hardware accelerators.