-
-
Notifications
You must be signed in to change notification settings - Fork 615
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
Removed dataflow sync from Engine #938
Conversation
- added deterministic methods - Fixed pytorch#818 - Fixed pytorch#795
pass | ||
|
||
|
||
def _repr_rng_state(rng_states): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why do we need this function ?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This can be helpful to debug random state
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Huge and awesome work :)
Engine is simplier, that's nice. IMO, do not deal with local seeds but with global and uniform strategy is a good way !
* Removed dataflow sync from Engine (#938) - added deterministic methods - Fixed #818 - Fixed #795 * Added DeterministicEngine (#939) * Fixed broken GPU test * Update examples and docs (#957) * Added DeterministicEngine * [WIP] Update examples * Fixes #941 for deterministic engine - added tests for both engines * Update examples/mnist/mnist_save_resume_engine * Updated mnist example, fixed minor bug in deterministic * autopep8 fix * Updated cifar10 docs + minor cosmetics * autopep8 fix * Cosmetics x * Commented results for single node, multi-gpus * Fixed mnist_save_resume_engine smoke tests Co-authored-by: AutoPEP8 <>
Description:
PR (1/3) to that splits #895
Removed dataflow sync from Engine
Added new engine events (used with deterministic methods)
DATALOADER_STOP_ITERATION : engine's specific event triggered when dataloader has no more data to provide
TERMINATE_SINGLE_EPOCH : triggered when the run is about to end the current epoch,
after receiving :meth:
~ignite.engine.Engine.terminate_epoch()
call.TERMINATE : triggered when the run is about to end completely,
after receiving :meth:
~ignite.engine.Engine.terminate()
call.Check list: