[FSDP][optim_state_dict] Enable cpu_offload config for optimzer state_dict #108434
Conversation
This file contains 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
…_dict We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) [ghstack-poisoned]
fegin
requested review from
mrshenli,
zhaojuanmao,
rohan-varma,
H-Huang,
awgu,
kwen2501,
wanchaol,
fduwjj,
wz337,
kiukchung and
d4l3k
as code owners
pytorch-bot
bot
added
the
release notes: distributed (fsdp)
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/108434
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (1 Unrelated Failure)As of commit dcd2e79 with merge base ada6550 ( UNSTABLE - The following job failed but was likely due to flakiness present on trunk and has been marked as unstable:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
fegin
added a commit
that referenced
this pull request
Sep 1, 2023
…_dict We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) ghstack-source-id: 199490000 Pull Request resolved: #108434
wz337
reviewed
Sep 2, 2023
…imzer state_dict" We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Sep 5, 2023
…_dict Pull Request resolved: #108434 We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. ghstack-source-id: 199739177 @exported-using-ghexport Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/)
…imzer state_dict" We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Sep 12, 2023
…_dict Pull Request resolved: #108434 We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. ghstack-source-id: 200499422 @exported-using-ghexport Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/)
…imzer state_dict" We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Sep 21, 2023
…_dict Pull Request resolved: #108434 We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. ghstack-source-id: 201500974 @exported-using-ghexport Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/)
…imzer state_dict" We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Oct 2, 2023
…_dict Pull Request resolved: #108434 We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. ghstack-source-id: 202584125 @exported-using-ghexport Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/)
wz337
added
the
ciflow/periodic
…imzer state_dict" We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Oct 3, 2023
…_dict Pull Request resolved: #108434 We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. ghstack-source-id: 202684640 @exported-using-ghexport Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/)
…imzer state_dict" We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Oct 3, 2023
…_dict Pull Request resolved: #108434 We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. ghstack-source-id: 202733483 @exported-using-ghexport Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/)
…imzer state_dict" We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Oct 3, 2023
…_dict Pull Request resolved: #108434 We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. ghstack-source-id: 202787456 @exported-using-ghexport Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/)
…imzer state_dict" We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/) [ghstack-poisoned]
fegin
added a commit
that referenced
this pull request
Oct 6, 2023
…_dict Pull Request resolved: #108434 We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag. ghstack-source-id: 203234325 @exported-using-ghexport Differential Revision: [D48913340](https://our.internmc.facebook.com/intern/diff/D48913340/)
|
@pytorchbot merge |
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):
We had the option but never used cpu_offload as optimizer state_dict offloads the tensors to CPU by default. And this is usually most users want as the tensors are required to be moved to CPU eventually. However, we may want to disable offloading to CPU in some cases, epsecially for the debugging purpose. This PR lets optimizer state_dict read the flag.
Differential Revision: D48913340