Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

How to adjust learning rate when the number of workers incresing? #27

Closed
Lihengwannafly opened this issue Dec 31, 2021 · 1 comment
Closed

Comments

@Lihengwannafly
Copy link

For example, I set original learning rate 0.001 when standalone mode, and 0.001 / sqrt(10) performs well when 10 workers running. But when 20 workers running, 0.001 / sqrt(20) performs very bad. So, is there any suggestion to adjust when the number of workers incresing?

@liutongxuan
Copy link
Member

liutongxuan commented Dec 31, 2021

There's list of topics about how to adjust the learning rate when increase total batch size in both sync mode or async mode.

This is an invalid, closed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants