[dynamo] add torch.compiler.set_stance #137504
Closed
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
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/137504
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 6248c56 with merge base 370d66d ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
williamwen42
commented
Oct 8, 2024
Attempt #2 at #132926 to implement #123771. Implement a new `torch.compiler.set_phase` function that can force `torch.compile` regions to run eagerly. Other options (such as "fail_on_recompile" and "eager_on_recompile") to be added in a followup PR. See added tests for usage examples. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
williamwen42
commented
Oct 8, 2024
Attempt #2 at #132926 to implement #123771. Implement a new `torch.compiler.set_phase` function that can force `torch.compile` regions to run eagerly. Other options (such as "fail_on_recompile" and "eager_on_recompile") to be added in a followup PR. See added tests for usage examples. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
yf225
reviewed
Oct 9, 2024
yf225
approved these changes
Oct 9, 2024
jansel
requested changes
Oct 10, 2024
|
Also note that a version of Similarly, |
Attempt #2 at #132926 to implement #123771. Implement a new `torch.compiler.set_phase` function that can force `torch.compile` regions to run eagerly. Other options (such as "fail_on_recompile" and "eager_on_recompile") to be added in a followup PR. See added tests for usage examples. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
williamwen42
changed the title
Attempt #2 at #132926 to implement #123771. Implement a new `torch.compiler.set_stance` function that can force `torch.compile` regions to run eagerly. See added tests for usage examples. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
|
I'm going to defer to @jansel on this since he has more opinions on the internal implementation details, but holler if you need me to unblock you |
jansel
requested changes
Oct 13, 2024
Attempt # 2 at #132926 to implement #123771. Implement a new `torch.compiler.set_stance` function that can force `torch.compile` regions to run eagerly. See added tests for usage examples. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
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 |
|
The merge job was canceled or timed out. This most often happen if two merge requests were issued for the same PR, or if merge job was waiting for more than 6 hours for tests to finish. In later case, please do not hesitate to reissue the merge command |
|
@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 |
yf225
added a commit
that referenced
this pull request
Oct 16, 2024
…to fallback to eager" Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
pytorchmergebot
pushed a commit
that referenced
this pull request
Oct 17, 2024
… to eager (#138113) Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". Pull Request resolved: #138113 Approved by: https://github.com/xmfan ghstack dependencies: #138105
|
Are we going to get the non-context manager directly mutating version of the API? |
|
@ezyang it's already implemented in this PR, although I overlooked adding a test and including it as a usage example. I can add this in a followup. |
yf225
added a commit
that referenced
this pull request
Oct 17, 2024
… decide whether to fallback to eager" Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
yf225
added a commit
that referenced
this pull request
Oct 17, 2024
…to fallback to eager" Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
|
@williamwen42 also I've started using the direct-mutating API: https://github.com/pytorch/pytorch/pull/138113/files and it's awesome :) |
yf225
added a commit
that referenced
this pull request
Oct 17, 2024
… decide whether to fallback to eager" Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
yf225
added a commit
that referenced
this pull request
Oct 17, 2024
…to fallback to eager" Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
yf225
added a commit
that referenced
this pull request
Oct 17, 2024
… decide whether to fallback to eager" Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
yf225
added a commit
that referenced
this pull request
Oct 17, 2024
…to fallback to eager" Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
yf225
added a commit
that referenced
this pull request
Oct 17, 2024
… decide whether to fallback to eager" Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
yf225
added a commit
that referenced
this pull request
Oct 17, 2024
…to fallback to eager" Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
pytorchmergebot
pushed a commit
that referenced
this pull request
Oct 18, 2024
… to eager (#138113) Dynamo stance is recently added in #137504. When Dynamo stance is "force_eager" (user explicitly wants to fall back to eager), we would like Compiled Autograd to fall back to eager as well. This will allow the Traceable FSDP2 use case to work since "eager forward + compiled autograd backward" is not supported for Traceable FSDP2. In general, if user wants to do "eager forward + compiled autograd backward", they should explicitly run the forward in eager instead of applying compile and then set stance to "force_eager". Pull Request resolved: #138113 Approved by: https://github.com/xmfan
pytorchmergebot
pushed a commit
that referenced
this pull request
Oct 18, 2024
…le (#138276) Followup to #137504 (comment) Pull Request resolved: #138276 Approved by: https://github.com/anijain2305, https://github.com/jansel
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
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):
Attempt # 2 at #132926 to implement #123771.
Implement a new
torch.compiler.set_stancefunction that can forcetorch.compileregions to run eagerly.See added tests for usage examples.
cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @chenyang78 @kadeng @chauhang @amjames @rec