-
Notifications
You must be signed in to change notification settings - Fork 25.6k
Automatically turn on dynamo in cond #108028
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
Conversation
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/108028
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 728b269 with merge base 1fd4e78 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
cc voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]
cc voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]
# We need to turn off the is_fx_tracing_flag. Remove this flag check from dyanmo | ||
# once we are confident fx tracing works with dynamo. | ||
torch.fx._symbolic_trace._is_fx_tracing_flag = False |
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.
Sorry, I still don't understand why we are turning this flag to False. You mentioned that making it go to True would make a lot of tests fail. But we shouldn't be regressing behavior, unless they're related to the symbool thing?
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.
make_fx will 1. turn this flag on and 2. feed FakeTensors and SymBools into torch.compile. I kind of assume this flag can be turned off in cond_compiled and it's not a problem.
Right now, if we leave this flag alone, we'll get an error for all make_fx tests, which isn't what we want we already supports "real" tracing mode.
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.
Is the problem that in torch.compile, we want _is_fx_tracing_flag
to be False? If so, I would have expected that we turned that off inside of torch.compile, because this is a make_fx and torch.compile composability issue, rather than doing a one-off solution here
|
||
self.assertTrue(torch.allclose(m(x), gm(x))) | ||
|
||
@unittest.expectedFailure |
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.
What's going on here? IIRC this test is important for quantization for 2.1?
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.
I'll have a check and get back. It has some complicated errors as I remember.
A replacement of #107932. cc voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]
A replacement of #107932. cc voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]
A replacement of #107932. cc voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]
A replacement of #107932. cc voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]
@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 |
Merge failedReason: 1 jobs have failed, first few of them are: trunk / macos-12-py3-arm64 / test (default, 3, 3, macos-m1-12) Details for Dev Infra teamRaised by workflow job |
A replacement of #107932. cc voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]
@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 |
Merge failedReason: 2 mandatory check(s) failed. The first few are:
Dig deeper by viewing the failures on hud |
A replacement of #107932. cc voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]
@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 |
Merge failedReason: Command
Details for Dev Infra teamRaised by workflow job |
@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):
A replacement of #107932.
cc @voznesenskym @penguinwu @anijain2305 @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @Xia-Weiwen @wenzhe-nrv @jiayisunx @chenyang78 @aakhundov