Add Scaled Dot Product Attention (SDPA) from PyTorch#5994
Merged
sw005320 merged 5 commits intoespnet:masterfrom Dec 28, 2024
Merged
Add Scaled Dot Product Attention (SDPA) from PyTorch#5994sw005320 merged 5 commits intoespnet:masterfrom
sw005320 merged 5 commits intoespnet:masterfrom
Conversation
for more information, see https://pre-commit.ci
Contributor
|
Thanks! |
for more information, see https://pre-commit.ci
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #5994 +/- ##
===========================================
+ Coverage 28.45% 56.08% +27.63%
===========================================
Files 829 830 +1
Lines 78052 78031 -21
===========================================
+ Hits 22208 43763 +21555
+ Misses 55844 34268 -21576
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
Contributor
|
Can you check some treatments for this library? |
Contributor
|
Thx! |
Shikhar-S
pushed a commit
to Shikhar-S/espnet
that referenced
this pull request
Mar 13, 2025
Add Scaled Dot Product Attention (SDPA) from PyTorch
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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?
This PR adds the SDPA implementation from PyTorch. It supports several implementations of the attention mechanism (including flash attention), some of which are more efficient than the default implementation.
See https://pytorch.org/docs/2.5/generated/torch.nn.functional.scaled_dot_product_attention.html for more information.
Note: SDPA does not seem to support flash attention with variable-length inputs. So, we still need to use the previous flash attention implementation.