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

DDP is incompatible with large datasets #1970

Closed
sabetAI opened this issue May 27, 2020 · 0 comments · Fixed by #2029
Closed

DDP is incompatible with large datasets #1970

sabetAI opened this issue May 27, 2020 · 0 comments · Fixed by #2029
Labels
help wanted Open to be worked on

Comments

@sabetAI
Copy link

sabetAI commented May 27, 2020 •
edited
Loading

I'm trying to stream a large data file instead of loading it so it doesn't have to be pickled for multi-processing. However, open-file objects give a TypeError: cannot serialize '_io.TextIOWrapper' object error, so I have to open it within a subprocess instead-- but train_dataloader and val_dataloader methods get called in the main process of pytorch-lightning! How can I bypass issue without changing the source code?
@sabetAI sabetAI added the help wanted Open to be worked on label May 27, 2020
@sabetAI sabetAI changed the title pytorch-lightning multi-processing is not compatible with large datasets pytorch-lightning multi-processing is incompatible with large datasets May 27, 2020
@sabetAI sabetAI changed the title pytorch-lightning multi-processing is incompatible with large datasets DDP is incompatible with large datasets May 27, 2020
@williamFalcon williamFalcon mentioned this issue May 31, 2020
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
help wanted Open to be worked on
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Replaces ddp .spawn with subprocess Lightning-AI/pytorch-lightning
1 participant
@sabetAI