refactor the total norm computation in grad clipping in APS #3243
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
meta-cla
bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
force-pushed
the
export-D79128843
branch
from
6d8d83c to
f57bf43
Compare
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 30, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the fsdp and ddp params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 30, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the fsdp and ddp params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
f57bf43 to
43738ef
Compare
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 30, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the fsdp and ddp params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
43738ef to
4f86531
Compare
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 30, 2025
…3243) Summary: Pull Request resolved: pytorch#3243 Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the fsdp and ddp params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
4f86531 to
f9be4b0
Compare
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 31, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the fsdp and ddp params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
f9be4b0 to
48c4acb
Compare
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 31, 2025
…3243) Summary: Pull Request resolved: pytorch#3243 Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the fsdp and ddp params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
48c4acb to
1305e2f
Compare
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 31, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the fsdp and ddp params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
1305e2f to
5203bf7
Compare
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 31, 2025
…3243) Summary: Pull Request resolved: pytorch#3243 Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
2 times, most recently
from
29f3764 to
5199ed0
Compare
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 31, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
force-pushed
the
export-D79128843
branch
from
5199ed0 to
d35993b
Compare
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Jul 31, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 4, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
4158816 to
e64c742
Compare
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
force-pushed
the
export-D79128843
branch
from
e64c742 to
499b0db
Compare
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 4, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 4, 2025
…3243) Summary: Pull Request resolved: pytorch#3243 Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
2 times, most recently
from
4ea53e1 to
43c4f99
Compare
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 4, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 4, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
43c4f99 to
98ebab3
Compare
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 5, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
2 times, most recently
from
969fda9 to
d7adb15
Compare
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 5, 2025
…3243) Summary: Pull Request resolved: pytorch#3243 Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 5, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
d7adb15 to
8fbd6ad
Compare
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 5, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
8fbd6ad to
edb808e
Compare
jialun-zhang
pushed a commit
to jialun-zhang/torchrec
that referenced
this pull request
Aug 5, 2025
…3243) Summary: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
jialun-zhang
force-pushed
the
export-D79128843
branch
from
edb808e to
39608e9
Compare
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
…3243) Summary: Pull Request resolved: pytorch#3243 Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the replicated and sharded params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter. Differential Revision: D79128843
|
This pull request was exported from Phabricator. Differential Revision: D79128843 |
jialun-zhang
force-pushed
the
export-D79128843
branch
from
39608e9 to
7e33149
Compare
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: Refactored the previous code for applying gradient clipping across ddp and fsdp parameter. Added a new funciton _compute_total_norm() that takes in the fsdp and ddp params provided in the gradientclippingOpitmizer class and computes the total gradient norm of the given parameter.
Differential Revision: D79128843