Feature/auto batch size find #426
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
benjijamorris
requested review from
hughes036,
saeliddp,
ritvikvasan and
yrkim98
hughes036
reviewed
Sep 27, 2024
| @@ -93,6 +100,10 @@ def train(cfg: DictConfig, data=None) -> Tuple[dict, dict]: | |||
| log.info("Logging hyperparameters!") | |||
| utils.log_hyperparameters(object_dict) | |||
|
|
|||
| if use_batch_tuner: | |||
There was a problem hiding this comment.
This seems related to the new block up starting on line 62 - could it go up there for readability? Could they also be a single condition (it seems they are logically bound?).
There was a problem hiding this comment.
Good point. I have it split up because the data arguments need to be changed if the batch size tuner is used, but then the actual batch tuning requires an initialized datamodule. I can rearrange things to make them a little closer, but that might be purely aesthetic as they can't be fully grouped.
hughes036
approved these changes
Sep 27, 2024
|
No new kwarg is added - just changing the default value of an existing kwarg. |
ritvikvasan
approved these changes
Oct 2, 2024
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Outdated suggestions cannot be applied.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.
What does this PR do?
The auto batch size finder increases the batch size by powers of 2 (starting with batch-size=1) by setting the datamodule's
batch_sizeattribute. This PR allows just the dataframe datamodule (I think the only one used by the plugin) to respond to its batch_size attribute being changed. In order to do this, we store the initial batch size (set to 1) and check for changes to the batch size to update the train dataloaders.Before submitting
pytestcommand?pre-commit run -acommand?Did you have fun?
Make sure you had fun coding 🙃