[pytorch] add manual registration for trace type by ljk53 · Pull Request #40903 · pytorch/pytorch

ljk53 · 2020-07-02T08:45:24Z

Stack from ghstack:

[pytorch] deprecate PYTORCH_DISABLE_TRACING macro #41004 [pytorch] deprecate PYTORCH_DISABLE_TRACING macro
[pytorch] add manual registration for trace type #40903 [pytorch] add manual registration for trace type

This PR continues the work of #38467 - decoupling Autograd and Trace for manually registered ops.

Differential Revision: D22354804

NOTE FOR REVIEWERS: This PR has internal Facebook specific changes or comments, please review them on Phabricator!

This PR continues the work of #38467 - decoupling Autograd and Trace for manually registered ops. Differential Revision: [D22354804](https://our.internmc.facebook.com/intern/diff/D22354804/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D22354804/)! [ghstack-poisoned]

dr-ci · 2020-07-02T08:56:29Z

💊 CI failures summary and remediations

As of commit 1790bba (more details on the Dr. CI page):

1/1 failures introduced in this PR

🕵️ 1 new failure recognized by patterns

The following CI failures do not appear to be due to upstream breakages:

pytorch_linux_backward_compatibility_check_test (1/1)

Step: "Run tests" (full log | diagnosis details | 🔁 rerun)

Jul 06 09:09:26 The PR is introducing backward incompatible changes to the operator library. Please contact PyTorch team to confirm whether this change is wanted or not.

Jul 06 09:09:26 processing existing schema:  __str__(__torch__.torch.classes._TorchScriptTesting._StackString _0) -> (str _0) 
Jul 06 09:09:26 processing existing schema:  __init__(__torch__.torch.classes._TorchScriptTesting._PickleTester _0, int[] _1) -> (None _0) 
Jul 06 09:09:26 processing existing schema:  __getstate__(__torch__.torch.classes._TorchScriptTesting._PickleTester _0) -> (int[] _0) 
Jul 06 09:09:26 processing existing schema:  __setstate__(__torch__.torch.classes._TorchScriptTesting._PickleTester _0, int[] _1) -> (None _0) 
Jul 06 09:09:26 processing existing schema:  top(__torch__.torch.classes._TorchScriptTesting._PickleTester _0) -> (int _0) 
Jul 06 09:09:26 processing existing schema:  pop(__torch__.torch.classes._TorchScriptTesting._PickleTester _0) -> (int _0) 
Jul 06 09:09:26 processing existing schema:  get(__torch__.torch.classes._TorchScriptTesting._LiteInterpreterTest _0, Tensor _1) -> (str _0) 
Jul 06 09:09:26 processing existing schema:  __getstate__(__torch__.torch.classes._TorchScriptTesting._LiteInterpreterTest _0) -> (int _0) 
Jul 06 09:09:26 processing existing schema:  __setstate__(__torch__.torch.classes._TorchScriptTesting._LiteInterpreterTest _0, int _1) -> (None _0) 
Jul 06 09:09:26 processing existing schema:  __init__(__torch__.torch.classes.dist_rpc.WorkerInfo _0, str _1, int _2) -> (None _0) 
Jul 06 09:09:26 The PR is introducing backward incompatible changes to the operator library. Please contact PyTorch team to confirm whether this change is wanted or not.  
Jul 06 09:09:26  
Jul 06 09:09:26 Broken ops: [ 
Jul 06 09:09:26 	aten::ne(str a, str b) -> (bool) 
Jul 06 09:09:26 	aten::eq(str a, str b) -> (bool) 
Jul 06 09:09:26 	aten::add(str a, str b) -> (str) 
Jul 06 09:09:26 ] 
Jul 06 09:09:26 =================== sccache compilation log =================== 
Jul 06 09:09:26 + cleanup 
Jul 06 09:09:26 + retcode=1 
Jul 06 09:09:26 + set +x

This comment was automatically generated by Dr. CI (expand for details).

Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions on the GitHub issue tracker or post in the (internal) Dr. CI Users group.

See how this bot performed.

This comment has been revised 13 times.

This PR continues the work of #38467 - decoupling Autograd and Trace for manually registered ops. Differential Revision: [D22354804](https://our.internmc.facebook.com/intern/diff/D22354804/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D22354804/)! [ghstack-poisoned]

Pull Request resolved: #40903 This PR continues the work of #38467 - decoupling Autograd and Trace for manually registered ops. ghstack-source-id: 107042483 Differential Revision: [D22354804](https://our.internmc.facebook.com/intern/diff/D22354804/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D22354804/)!

This PR continues the work of #38467 - decoupling Autograd and Trace for manually registered ops. Differential Revision: [D22354804](https://our.internmc.facebook.com/intern/diff/D22354804/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D22354804/)! [ghstack-poisoned]

Pull Request resolved: #40903 This PR continues the work of #38467 - decoupling Autograd and Trace for manually registered ops. ghstack-source-id: 107087630 Differential Revision: [D22354804](https://our.internmc.facebook.com/intern/diff/D22354804/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D22354804/)!

This PR continues the work of #38467 - decoupling Autograd and Trace for manually registered ops. Differential Revision: [D22354804](https://our.internmc.facebook.com/intern/diff/D22354804/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D22354804/)! [ghstack-poisoned]

Pull Request resolved: #40903 This PR continues the work of #38467 - decoupling Autograd and Trace for manually registered ops. ghstack-source-id: 107106964 Differential Revision: [D22354804](https://our.internmc.facebook.com/intern/diff/D22354804/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D22354804/)!

This PR continues the work of #38467 - decoupling Autograd and Trace for manually registered ops. Differential Revision: [D22354804](https://our.internmc.facebook.com/intern/diff/D22354804/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D22354804/)! [ghstack-poisoned]

ezyang · 2020-07-06T18:08:41Z

This is good stuff, but I think there's a decent chance the tracing logic here didn't need to be manually written out. Looking at the original manual code, you can see that there are bits that were codegenned, and then bits that were edited up to do some sort of custom logic. In most of the cases, the tracing logic was outside of the "this was manually edited sections." How hard would it be to wire up the codegen to generate tracing for those cases too?

(Maybe this is not worth it, since the PoR is to shortly delete all this code and replace it with boxed fallback.)

ezyang · 2020-07-06T18:34:20Z

        x = torch.randn(4, 4, requires_grad=True)

        def f(x):
-            out = Variable(torch.zeros(x.size()))


ezyang

There is not too much manual code that was carried over, so I think this is fine as is to go in.

ljk53 · 2020-07-06T18:47:21Z

There is not too much manual code that was carried over, so I think this is fine as is to go in.

Yeah, most ops here are simply falling through, among all the 5 manually registered ops: copy_ has special logic to emit as expand_as, resize_ and resize_as_ only to emit warning message (not sure if it's still right thing to do?), only detach and detach_ are regular form that can be replaced by codegen - but we'll need create a separate boolean flag to differentiate these 2 from other manual registered ops, so I decided not to do so...

facebook-github-bot · 2020-07-06T22:11:04Z

This pull request has been merged in 7f60642.

Summary: Pull Request resolved: pytorch#40903 This PR continues the work of pytorch#38467 - decoupling Autograd and Trace for manually registered ops. ghstack-source-id: 107158638 Test Plan: CI Differential Revision: D22354804 fbshipit-source-id: f5ea45ade2850296c62707a2a4449d7d67a9f5b5

ljk53 requested review from albanD and apaszke as code owners July 2, 2020 08:45

ljk53 requested review from bhosmer, ezyang and smessmer July 2, 2020 10:32

albanD reviewed Jul 2, 2020

View reviewed changes

Comment thread test/jit/test_tracer.py Outdated

Comment thread torch/csrc/autograd/TraceTypeManual.cpp

ljk53 mentioned this pull request Jul 6, 2020

[pytorch] deprecate PYTORCH_DISABLE_TRACING macro #41004

Closed

ezyang reviewed Jul 6, 2020

View reviewed changes

ezyang approved these changes Jul 6, 2020

View reviewed changes

facebook-github-bot closed this in 7f60642 Jul 6, 2020

facebook-github-bot added the merged label Jul 6, 2020

facebook-github-bot deleted the gh/ljk53/154/head branch July 10, 2020 14:18

mruberry added the Merged label Oct 28, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[pytorch] add manual registration for trace type#40903

[pytorch] add manual registration for trace type#40903
ljk53 wants to merge 6 commits intogh/ljk53/154/basefrom
gh/ljk53/154/head

ljk53 commented Jul 2, 2020 •

edited

Loading

Uh oh!

dr-ci Bot commented Jul 2, 2020 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

ezyang commented Jul 6, 2020

Uh oh!

ezyang Jul 6, 2020

Uh oh!

ezyang left a comment

Uh oh!

ljk53 commented Jul 6, 2020

Uh oh!

facebook-github-bot commented Jul 6, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Conversation

ljk53 commented Jul 2, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dr-ci Bot commented Jul 2, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

💊 CI failures summary and remediations

🕵️ 1 new failure recognized by patterns

pytorch_linux_backward_compatibility_check_test (1/1)

Uh oh!

Uh oh!

Uh oh!

ezyang commented Jul 6, 2020

Uh oh!

ezyang Jul 6, 2020

Choose a reason for hiding this comment

Uh oh!

ezyang left a comment

Choose a reason for hiding this comment

Uh oh!

ljk53 commented Jul 6, 2020

Uh oh!

facebook-github-bot commented Jul 6, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

ljk53 commented Jul 2, 2020 •

edited

Loading

dr-ci Bot commented Jul 2, 2020 •

edited

Loading