validation and training loops run the partial dataset #1192
Conversation
This file contains hidden or 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
Borda
changed the title
Contributor
|
@srush mind taking a look? this came from our chats with the HF code. |
Contributor
|
This seems good to me. We have some val sets that are quite large. |
alexeykarnachev
pushed a commit
to alexeykarnachev/pytorch-lightning
that referenced
this pull request
Apr 3, 2020
) * auto_add_sampler() fix * auto_add_sampler() fix Co-authored-by: seth <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
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Outdated suggestions cannot be applied.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.
Fix #1161 - when using ddp/ddpd2, the validation and training loops run the full respective dataset on each gpu. This costs time, and changes batch counts for any statistics being collected.
The fix just makes sure that for ddp and ddp2,
auto_add_sampler()creates aDistributedSamplerfor each data set.This passes all the tests on my machine except for slurm and apex related as I do not have either. I don't think this needs any doc changes. I can look into writing a test for this ... if needed. Let me know.