Fix gradient shape error for DPMultiheadAttention (issue 650) #651
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
facebook-github-bot
added
the
CLA Signed
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D57446245 |
HuanyuZhang
added a commit
to HuanyuZhang/opacus
that referenced
this pull request
May 16, 2024
…ytorch#651) Summary: When batch_first = True, the activation and partial gradient for each linear layer in DPMultiheadAttention still has batch_size in the second dimension, thus causing wrong gradient shape in optimizer.step(). Details in: meta-pytorch#650 Differential Revision: D57446245
HuanyuZhang
force-pushed
the
export-D57446245
branch
from
cbc61b8 to
0d8ccfd
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D57446245 |
HuanyuZhang
added a commit
to HuanyuZhang/opacus
that referenced
this pull request
May 30, 2024
…ytorch#651) Summary: When batch_first = True, the activation and partial gradient for each linear layer in DPMultiheadAttention still has batch_size in the second dimension, thus causing wrong gradient shape in optimizer.step(). Details in: meta-pytorch#650 Differential Revision: D57446245
HuanyuZhang
force-pushed
the
export-D57446245
branch
from
0d8ccfd to
98af8b9
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D57446245 |
…ytorch#651) Summary: When batch_first = True, the activation and partial gradient for each linear layer in DPMultiheadAttention still has batch_size in the second dimension, thus causing wrong gradient shape in optimizer.step(). Details in: meta-pytorch#650 Reviewed By: EnayatUllah Differential Revision: D57446245
HuanyuZhang
force-pushed
the
export-D57446245
branch
from
98af8b9 to
d4bc8c7
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D57446245 |
Contributor
|
This pull request has been merged in 202c58a. |
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.
Summary:
When batch_first = True, the activation and partial gradient for each linear layer in DPMultiheadAttention still has batch_size in the second dimension, thus causing wrong gradient shape in optimizer.step().
Details in: #650
Differential Revision: D57446245