Merged
Conversation
|
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. |
dacorvo
force-pushed
the
remove_unsupported_modeling_flags
branch
from
2265176 to
63110ce
Compare
This kernel is not supported on Trainium 1, and its integration is likely to change anyway when we support Trainium 2.
This kernel is not supported on Trainium 1, and its integration is likely to change anyway when we support Trainium 2.
dacorvo
force-pushed
the
remove_unsupported_modeling_flags
branch
from
63110ce to
a13a4a2
Compare
Note that flash attention is still used for prefill whenever it is relevant.
It is never set to something else than 2, and it can be overridden in the model compiler flags anyway.
Collaborator
Author
|
@tengomucho thank you for the review, but I realized that while removing flash decoding I had also inadvertently removed flash attention for prefill. Will force-push. |
dacorvo
force-pushed
the
remove_unsupported_modeling_flags
branch
from
a13a4a2 to
3523b40
Compare
tengomucho
reviewed
Sep 5, 2025
|
|
||
| return FlashAttentionStrategy.NONE | ||
|
|
||
| def compute_for_flash_decoding(self, Q, K, V, past_key_value, attention_mask, active_mask) -> Tensor: |
Collaborator
There was a problem hiding this comment.
I see, you need to restore these bits
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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?
This removes some optimized code paths that are not supported with the current models (mainly Llama and its variants) and/or deployment configurations (Trainium instances).