More flexible RNG synchronization #8
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
LysandreJik
approved these changes
Mar 9, 2021
| @@ -42,6 +44,18 @@ class Accelerator: | |||
| cpu (:obj:`bool`, `optional`): | |||
| Whether or not to force the script to execute on CPU. Will ignore GPU available if set to :obj:`True` and | |||
| force the execution on one process only. | |||
| rng_types (list of :obj:`str` or :class:`~accelerate.utils.RNGType`): | |||
src/accelerate/utils.py
Outdated
| @@ -35,28 +44,47 @@ def set_seed(seed: int): | |||
| np.random.seed(seed) | |||
| torch.manual_seed(seed) | |||
| torch.cuda.manual_seed_all(seed) | |||
| if is_tpu_available(): | |||
| xm.set_rng_state(seed) | |||
| # ^^ safe to call this function even if cuda is not available | |||
There was a problem hiding this comment.
Maybe you can move this comment above if you want to keep it
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently, the RNG synchronization at the beginning of each epoch is done very broadly, which means the potential data augmentation is done the same way on each process. While this is probably not hurting a big deal unless the underlying dataset isn't very big, there is a way to use local RNG with
torch.Generatorintroduced in PyTorch 1.6.This PR enables this by default and leaves some flexibility to the user to set the RNGs they want to synchronize (among torch, cuda, xla and local generator).