Neg n_workers are now the same as zero #4019
Merged
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
|
FWIW, scikit-learn and joblib use a different convention: -1 means "use the same number of workers as CPU cores", -2 means "use num_cores-1", etc. From joblib docs:
|
|
Taking into account @kmike's comment I feel like it'd be best to raise |
|
Changed it to a |
apaszke
approved these changes
Dec 5, 2017
apaszke
reviewed
Dec 5, 2017
torch/utils/data/dataloader.py
Outdated
| # prime the prefetch loop | ||
| for _ in range(2 * self.num_workers): | ||
| self._put_indices() | ||
| elif self.num_workers < 0: |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
Erotemic
force-pushed
the
fix_neg_num_workers
branch
from
1675c26 to
cb44b81
Compare
apaszke
approved these changes
Dec 5, 2017
|
Thank you! |
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Outdated suggestions cannot be applied.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.
I foolishly passed the kwarg
num_workers=-1to aDataLoaderexpecting that it would default to serial processing (I correctly recalled that it should be a non-positive number, but failed to remember that it must be zero to get this behavior).Normally this would be fine, but the error message that it spat out when I did this was very confusing:
What made the matter worse is that I recently put a print statement in that file to test something, so I though I had broken pytorch, but I couldn't figure out where the difference was. After spending too much time searching for cached pyc files, I finally realized my mistake.
I wanted to submit a patch so it would at least spit out a ValueError. However, I think it is a simpler change to have it accept negative numbers. This change simply makes it so setting num_workers to a non-positive number results in the same behavior as zero.