Skip to content

[PyTorch Debug] Fix the issue with PP#1894

Merged
pggPL merged 4 commits intoNVIDIA:mainfrom
pggPL:pp_nvinspect
Jun 26, 2025
Merged

[PyTorch Debug] Fix the issue with PP#1894
pggPL merged 4 commits intoNVIDIA:mainfrom
pggPL:pp_nvinspect

Conversation

@pggPL
Copy link
Copy Markdown
Collaborator

@pggPL pggPL commented Jun 25, 2025

Description

Simple fix for PP.

Fixes # (issue)

Type of change

  • Documentation change (change only to the documentation, either a fix or a new content)
  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • Infra/Build change
  • Code refactoring

Changes

Please list the changes introduced in this PR:

  • Change A
  • Change B

Checklist:

  • I have read and followed the contributing guidelines
  • The functionality is complete
  • I have commented my code, particularly in hard-to-understand areas
  • I have made corresponding changes to the documentation
  • My changes generate no new warnings
  • I have added tests that prove my fix is effective or that my feature works
  • New and existing unit tests pass locally with my changes

pggPL and others added 4 commits June 25, 2025 10:58
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>
@pggPL
Copy link
Copy Markdown
Collaborator Author

pggPL commented Jun 25, 2025

/te-ci pytorch

@lengerfulluse
Copy link
Copy Markdown

# .internal = True is slightly faster, but results
# in errors when caching the weights.

@pggPL thanks for working on this. The PP > 1 verified worked when i set to False. But wondering whether you have some metrics on the throughput impact on this when you commented with slightly faster

@pggPL pggPL merged commit 964c2ed into NVIDIA:main Jun 26, 2025
19 of 21 checks passed
@pggPL
Copy link
Copy Markdown
Collaborator Author

pggPL commented Jun 26, 2025

@lengerfulluse we use internal=False to mitigate CPU overhead with creating tensor object. It should not affect layers in iterations without logging and the overhead for logged layers should be much smaller than other overhead of logging.

KshitijLakhani pushed a commit that referenced this pull request Jun 27, 2025
* fix

Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix

Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
chtruong814 pushed a commit to chtruong814/TransformerEngine that referenced this pull request Jul 17, 2025
* fix

Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix

Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants