Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[PyTorch] Fix backward compatibility for checkpoint API #748

Merged
merged 3 commits into from
Apr 4, 2024
Merged

[PyTorch] Fix backward compatibility for checkpoint API #748

merged 3 commits into from
Apr 4, 2024

Conversation

ksivaman
Copy link
Member

@ksivaman ksivaman commented Apr 4, 2024

A follow-up to #740

@ksivaman
Args can be None
d13f75c 
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
@ksivaman ksivaman added the 1.5.0 label Apr 4, 2024
@ksivaman ksivaman requested a review from denera April 4, 2024 00:54
@ksivaman ksivaman self-assigned this Apr 4, 2024
@ksivaman
Copy link
Member Author

ksivaman commented Apr 4, 2024

/te-ci pytorch

@erhoo82
Copy link
Collaborator

erhoo82 commented Apr 4, 2024 •
edited
Loading

I tested this with both pretraining, LoRA, and SFT.
In the case of LoRA and SFT, this is one of the input types passed as arg, which is not captured.
https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/core/packed_seq_params.py#L7

ksivaman added 2 commits April 3, 2024 19:35
@ksivaman
Merge branch 'main' into fix_backward_compatibility
b5e99e7
@ksivaman
Fix other arg types
77b4fd3 
Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
@ksivaman
Copy link
Member Author

ksivaman commented Apr 4, 2024

/te-ci pytorch

@ksivaman ksivaman merged commit ce328fa into NVIDIA:main Apr 4, 2024
9 of 10 checks passed
ksivaman added a commit that referenced this pull request Apr 4, 2024
@ksivaman
[PyTorch] Fix backward compatibility for checkpoint API (#748)
6a9edc3 
* Args can be None

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix other arg types

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
pggPL pushed a commit to pggPL/TransformerEngine that referenced this pull request May 15, 2024
@ksivaman @pggPL
[PyTorch] Fix backward compatibility for checkpoint API (NVIDIA#748)
3e2cca8 
* Args can be None

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix other arg types

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>
pggPL pushed a commit to pggPL/TransformerEngine that referenced this pull request May 23, 2024
@ksivaman @pggPL
[PyTorch] Fix backward compatibility for checkpoint API (NVIDIA#748)
6338367 
* Args can be None

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

* Fix other arg types

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>

---------

Signed-off-by: Kirthi Shankar Sivamani <ksivamani@nvidia.com>
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@denera denera denera approved these changes

Assignees

@ksivaman ksivaman

Labels
1.5.0
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@ksivaman @erhoo82 @denera