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.

Sign up for GitHub

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

Jump to bottom

Regarding the batch size used for different model training #5376

Closed
Zengyf-CVer opened this issue Oct 28, 2021 · 4 comments
Closed

Regarding the batch size used for different model training #5376

Zengyf-CVer opened this issue Oct 28, 2021 · 4 comments
Labels
question Further information is requested

Comments

@Zengyf-CVer
Copy link
Contributor

Zengyf-CVer commented Oct 28, 2021 •
edited
Loading

@glenn-jocher
I saw in README that you set batch-size like this. As far as I know, batch-size is generally 2 to the n-th power, but what is the setting of 40 and 24 here? Or is it arbitrarily set according to the different GPU memory usage?

ksnip_20211028-185405
@Zengyf-CVer Zengyf-CVer added the question Further information is requested label Oct 28, 2021
@glenn-jocher
Copy link
Member

@Zengyf-CVer arbitrary for single-GPU, or use --batch -1 for autobatch.

@Zengyf-CVer
Copy link
Contributor Author

@glenn-jocher
Is this autobatch designed based on the maximum memory of a single GPU?

@glenn-jocher
Copy link
Member

@Zengyf-CVer it's an experimental feature only available for single-GPU training at the moment. See #5092

@Zengyf-CVer
Copy link
Contributor Author

@glenn-jocher
Thank you very much.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
question Further information is requested
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@glenn-jocher @Zengyf-CVer