Towards making the interface of ghost clipping same as that of PyTorch #668
Conversation
This file contains 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
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 22, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
1ab9ab2
to
0b9ff65
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
force-pushed
the
export-D61162530
branch
from
0b9ff65
to
497d62d
Compare
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 22, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 22, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
497d62d
to
0225c49
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 22, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
0225c49
to
d6e05bf
Compare
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 22, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
d6e05bf
to
0b872c4
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
1 similar comment
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 22, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
0b872c4
to
8e20d3d
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 22, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
8e20d3d
to
0edb702
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 23, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
0edb702
to
6eb8da3
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 27, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
6eb8da3
to
3264374
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 28, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
3264374
to
03c692b
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 28, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
03c692b
to
a10cea0
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 28, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
a10cea0
to
5fcf4d6
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 28, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
5fcf4d6
to
214dcd8
Compare
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
added a commit
to EnayatUllah/opacus
that referenced
this pull request
Aug 28, 2024
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
EnayatUllah
force-pushed
the
export-D61162530
branch
from
214dcd8
to
710d2cb
Compare
pytorch#668) Summary: Pull Request resolved: pytorch#668 We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping. Differential Revision: D61162530
|
This pull request was exported from Phabricator. Differential Revision: D61162530 |
EnayatUllah
force-pushed
the
export-D61162530
branch
from
710d2cb
to
29a96b6
Compare
|
This pull request has been merged in 36f7a34. |
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.
Summary: We define two classes DPLossFastGradientClipping and DPTensorFastGradientClipping in the utils fine, which allows us to repurpose loss.backward() to perform the two backward passes and loss scaling required to implement ghost clipping.
Differential Revision: D61162530