ACiD.jl
This is an implementation of an asynchronous multiprocessing optimization algorithm with a continuous local momentum called A²CiD² in the Julia programming language as introduced in [1].
There is also an official demo by the original authors of [1] in Python: AdelNabli/ACiD. This code was also used as a reference for our implementation.
Installation
Resources
Index
- 1A. Nabli, E. Belilovsky, and E. Oyallon, “A²CiD²: Accelerating Asynchronous Communication in Decentralized Deep Learning,” in Thirty-seventh Conference on Neural Information Processing Systems, 2023. [Online]. Available:
