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

Reverse transform of state_dict per submodule #1011

Merged
merged 6 commits into from
Aug 21, 2024
Merged

Reverse transform of state_dict per submodule #1011

merged 6 commits into from
Aug 21, 2024

Conversation

crcrpar
Copy link
Collaborator

@crcrpar crcrpar commented Aug 21, 2024

What does this PR do?

Implements reverse transform to dump a state dict of unsharded parameters, for FSDP.

successor of #913

crcrpar added 5 commits August 21, 2024 21:15
@crcrpar
init
93d3a04 
Signed-off-by: Masaki Kozuki <mkozuki@nvidia.com>
@crcrpar
old shape is unsharded, new is sharded
3503a68 
Signed-off-by: Masaki Kozuki <mkozuki@nvidia.com>
@crcrpar
docstring
76c00f4 
Signed-off-by: Masaki Kozuki <mkozuki@nvidia.com>
@crcrpar
docstring
a70489c 
Signed-off-by: Masaki Kozuki <mkozuki@nvidia.com>
@crcrpar
remove unused member method
2ea8961 
Signed-off-by: Masaki Kozuki <mkozuki@nvidia.com>
t-vi
t-vi reviewed Aug 21, 2024
View reviewed changes
Copy link
Collaborator

@t-vi t-vi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Absolutely cool! Very excited to see this. Thank you @crcrpar
I have a few details about the return value, but other than this, it seems great from all sides.

thunder/core/module.py Outdated Show resolved Hide resolved
thunder/distributed/transforms/fsdp_v2.py Show resolved Hide resolved
thunder/distributed/transforms/fsdp_v2.py Show resolved Hide resolved
thunder/core/module.py Outdated Show resolved Hide resolved
thunder/tests/distributed/test_fsdp.py Show resolved Hide resolved
@crcrpar
destination update
2267a23 
Signed-off-by: Masaki Kozuki <mkozuki@nvidia.com>
@t-vi t-vi enabled auto-merge (squash) August 21, 2024 17:53
t-vi
t-vi approved these changes Aug 21, 2024
View reviewed changes
Copy link
Collaborator

@t-vi t-vi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Exiting new aspect of transforms. Thank you @crcrpar !

@t-vi t-vi merged commit c92e8a8 into main Aug 21, 2024
37 checks passed
@t-vi t-vi deleted the crpa/reverse_transform_state_dict branch August 21, 2024 17:55
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@t-vi t-vi t-vi approved these changes

@mruberry mruberry Awaiting requested review from mruberry mruberry is a code owner

@lantiga lantiga Awaiting requested review from lantiga lantiga is a code owner

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@crcrpar @t-vi