[Torch] Support dynamic head_dim in attention conversion#23680
Closed
keshavvinayak01 wants to merge 1 commit intoiree-org:mainfrom
Closed
[Torch] Support dynamic head_dim in attention conversion#23680keshavvinayak01 wants to merge 1 commit intoiree-org:mainfrom
keshavvinayak01 wants to merge 1 commit intoiree-org:mainfrom
Conversation
Remove the NYI error for dynamic head_dim and compute scale at runtime via math.rsqrt when the dimension is not statically known. Add lit tests for dynamic head_dim cases. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> Signed-off-by: Keshav Vinayak Jha <keshavvinayakjha@gmail.com>
Contributor
Author
|
Closing, duplicate of #23636 |
This file contains hidden or 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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
When
head_dimis dynamic, scale is computed at runtime viaindex_cast+sitofp+rsqrtinstead of being folded to a constant.Static
head_dimbehavior is unchanged.