40% Adam Optimizer Improvement
If you are training deep learning models in PyTorch, default optimizer implementations leave performance on the table.
Check out this project demonstrating up to a 40% Adam optimizer improvement:
https://github.com/yangofzeal/adam/
Highlights
- 40% faster step execution time in PyTorch.
- Drop-in replacement for standard Adam.
- Open-source implementation available on GitHub: https://github.com/yangofzeal/adam/
Top comments (0)