Change accumulate_train_batch_on_tokens default to True #1618
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
snarayan21
approved these changes
Oct 27, 2024
|
Open to other opinions, but given this is a correctness change, I think it is ok to change without deprecation (also deprecation warning for a default change is a bit unlikely to have any impact). Definitely will call this out in the next release notes. |
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Outdated suggestions cannot be applied.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.
Following #1610, update the default value of
accumulate_train_batch_on_tokenstoTrue, for a more mathematically correct default. Note: this will slightly change loss curves for models trained with padding. The old behavior can be recovered if desired. by simply setting this toFalseexplicitly. See #1610 and mosaicml/composer#3677 for more discussion of this change.