Improve argument validation for Flash-attn + SWA #1162
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
|
|
|
Ready for review! |
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
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Outdated suggestions cannot be applied.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.
As raised by @malteos , Flash attention requires not 2.0.0+ but version 2.3.0+ to support sliding_window_width. However also, sliding-window attention is supported in the absence of Flash.
This PR changes the argument validation to 1) check for version 2.3.0 and 2) only perform this check if "flash" type attention is used. Also, it doesn't pass window_size kwarg when not required, meaning we'll never give the invalid kwarg to flash versions < 2.3.0 .