[core] Refactor hub attn kernels #12475
Open
Conversation
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
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
MekkCyber
reviewed
Oct 13, 2025
Comment on lines
+467
to
+471
| def _prepare_attention_backend(backend: AttentionBackendName) -> None: | ||
| preparer = _BACKEND_PREPARERS.get(backend) | ||
| if preparer is not None: | ||
| preparer() | ||
|
|
There was a problem hiding this comment.
I'm not sure I understand what the preparer does
There was a problem hiding this comment.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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Outdated suggestions cannot be applied.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?
Refactors how we load the attention kernels from the Hub.
Currently, when a user specifies the
DIFFUSERS_ENABLE_HUB_KERNELSenv var, we always download the supported kernel. Currently, we have FA3, but we have ongoing PRs that support FA and SAGE: #12387 and #12439. So, we will download ALL of them even when they're not required. This is not good.This PR makes it so that only the relevant kernel gets downloaded without breaking
torch.compilecompliance (fullgraph and no recompilation triggers).Cc: @MekkCyber