❓🧐 Improve automatic memory optimization parameter handling #404
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
lvermue
changed the title
cthoyt
changed the title
cthoyt
reviewed
Apr 26, 2021
| @@ -1000,8 +1000,8 @@ def pipeline( # noqa: C901 | |||
| mapped_triples = validation.mapped_triples | |||
|
|
|||
| # Evaluate | |||
| # Reuse optimal evaluation parameters from training if available | |||
| if evaluator_instance.batch_size is not None or evaluator_instance.slice_size is not None: | |||
| # Reuse optimal evaluation parameters from training if available, only if the validation triples are used again | |||
There was a problem hiding this comment.
when this says training, it's because it runs a fake "training" with the mapped triples for the evaluation data?
lvermue
changed the title
|
@PyKEEN-bot Please test this! |
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Outdated suggestions cannot be applied.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 pipeline re-uses derived optimal parameter settings from the automatic memory optimization.
This can lead to issues, because these parameters are derived from using the validation dataset. In cases where the test dataset is bigger than the validation dataset, these re-used parameter are overly optimistic and can lead to crashes. Furthermore, in the light of #398 this issue is significantly increased.
Therefore, this PR will mitigate this issue by making the re-usage of the parameters depending on whether the validation dataset is also used for the final evaluation.