-
Notifications
You must be signed in to change notification settings - Fork 21.4k
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.
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
[FSDP][optim_state_dict] Ensure correct devices for tensors when doing all_gather #92992
Conversation
…g all_gather [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/92992
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 90022ac: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
…g all_gather ghstack-source-id: 3bb40dbbac0df7a2772e96ee5c1005ce227bf12b Pull Request resolved: #92992
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
lgtm thanks!
…s when doing all_gather" When doing `_all_gather_optim_state`, we need to ensure that `step` tensors are on CPU and other tensors are on GPUs. This PR add the logic to ensure the locality. [ghstack-poisoned]
…g all_gather ghstack-source-id: f9f72d8fbbce3a7a90b2c971cca760c0b47bf26c Pull Request resolved: #92992
…s when doing all_gather" When doing `_all_gather_optim_state`, we need to ensure that `step` tensors are on CPU and other tensors are on GPUs. This PR add the logic to ensure the locality. [ghstack-poisoned]
…g all_gather ghstack-source-id: b6c2b50393ff9935614b84df33c388847458463a Pull Request resolved: #92992
@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 |
Stack from ghstack (oldest at bottom):
When doing
_all_gather_optim_state
, we need to ensure thatstep
tensors are on CPU and other tensors are on GPUs. This PR add the logic to ensure the locality.