[FSDP][optim_state_dict] Make FSDP.optim_state_dict compatbile with DMP #93285
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
`torchrec.DistributedModelParallel` overwrites `named_parameters` and is not compatible with `FullyShardedDataParallel`'s optim_state_dict. This PR adds some workaround in `FullyShardedDataParallel` to make both work together. Differential Revision: [D42764611](https://our.internmc.facebook.com/intern/diff/D42764611/) [ghstack-poisoned]
fegin
requested review from
H-Huang,
awgu,
kwen2501,
mrshenli,
rohan-varma,
wanchaol and
zhaojuanmao
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/93285
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 8c9e355: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
the
release notes: distributed (fsdp)
fegin
added a commit
that referenced
this pull request
Jan 30, 2023
`torchrec.DistributedModelParallel` overwrites `named_parameters` and is not compatible with `FullyShardedDataParallel`'s optim_state_dict. This PR adds some workaround in `FullyShardedDataParallel` to make both work together. Differential Revision: [D42764611](https://our.internmc.facebook.com/intern/diff/D42764611/) ghstack-source-id: 178786341 Pull Request resolved: #93285
…bile with DMP" `torchrec.DistributedModelParallel` overwrites `named_parameters` and is not compatible with `FullyShardedDataParallel`'s optim_state_dict. This PR adds some workaround in `FullyShardedDataParallel` to make both work together. Differential Revision: [D42764611](https://our.internmc.facebook.com/intern/diff/D42764611/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Jan 31, 2023
Pull Request resolved: #93285 `torchrec.DistributedModelParallel` overwrites `named_parameters` and is not compatible with `FullyShardedDataParallel`'s optim_state_dict. This PR adds some workaround in `FullyShardedDataParallel` to make both work together. ghstack-source-id: 178905353 Differential Revision: [D42764611](https://our.internmc.facebook.com/intern/diff/D42764611/)
rohan-varma
approved these changes
Feb 1, 2023
| # overwite the flat_parameters traversal result to only obtain | ||
| # the last one, which happens to be the correct one. | ||
| # | ||
| # TODO: Remove this hack once DMP + FSDP is not supported. |
Contributor
There was a problem hiding this comment.
does this mean it should be removed once we've landed the composable path?
Contributor
Author
There was a problem hiding this comment.
If all existing use cases are migrated to the composable path, then yes, we should only support trech_shard + fully_shard but not DMP + FSDP.
…bile with DMP" `torchrec.DistributedModelParallel` overwrites `named_parameters` and is not compatible with `FullyShardedDataParallel`'s optim_state_dict. This PR adds some workaround in `FullyShardedDataParallel` to make both work together. Differential Revision: [D42764611](https://our.internmc.facebook.com/intern/diff/D42764611/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Feb 2, 2023
Pull Request resolved: #93285 `torchrec.DistributedModelParallel` overwrites `named_parameters` and is not compatible with `FullyShardedDataParallel`'s optim_state_dict. This PR adds some workaround in `FullyShardedDataParallel` to make both work together. ghstack-source-id: 179122189 Differential Revision: [D42764611](https://our.internmc.facebook.com/intern/diff/D42764611/)
Contributor
Author
|
@pytorchbot merge |
Collaborator
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Collaborator
|
The merge job was canceled. If you believe this is a mistake,then you can re trigger it through pytorch-bot. |
Contributor
Author
|
@pytorchbot merge |
Collaborator
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
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.
Stack from ghstack (oldest at bottom):
torchrec.DistributedModelParalleloverwritesnamed_parametersand is not compatible withFullyShardedDataParallel's optim_state_dict. This PR adds some workaround inFullyShardedDataParallelto make both work together.Differential Revision: D42764611