fix forward batching for seq2seq and right padding models. #139
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
|
The documentation is not available anymore as the PR was closed or merged. |
younesbelkada
approved these changes
Feb 9, 2023
There was a problem hiding this comment.
This looks great to me! 💯
[optional] My only comment:
- We can maybe not rely on
self.datacollatorand create theattention_maskourselves (for e.g.attention_mask = q.ne(self.tokenizer.pad_token_id) * 1.0) ? This way if a user wants to build something on top of that can directly see where the correctattention_maskis created. i.e. passing a full-ones todata_collatorto retrieve the correct one is IMO not very intuitive - maybe at least worth commenting it
|
Both runs seem to be converging normally: |
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Outdated suggestions cannot be applied.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.
This PR fixes issues with forward batching of models that pad to the right (e.g.
bloom) and seq2seq models (e.g.T5).