Skip to content

[Caffe2] Fix shape inference for element-wise operators #33431

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.

Sign up for GitHub

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

Jump to bottom
Closed
wants to merge 1 commit into from
Closed

[Caffe2] Fix shape inference for element-wise operators #33431

wants to merge 1 commit into from

Conversation

cheparukhin
Copy link

Summary:
Some elementwise operators don't have shape and type inference specified for the output tensor: BitwiseOr, BitwiseAnd, BitwiseXor, Not, Sign.

This change fixes this issue:

  • For Not and Sign operators, the output has the same type and shape as the input, so IdenticalTypeAndShapeOfInput function is used to specify that.
  • For bitwise operators created by CAFFE2_SCHEMA_FOR_BINARY_BITWISE_OP macro, the type and shape inference rules should be the same as for other binary element-wise operators, so TensorInferenceFunction(ElementwiseOpShapeInference) is used to specify that.

Also some tests were modified to ensure that the shape and type are inferred (ensure_outputs_are_inferred parameter)

Test Plan:

CAFFE2_ASSERT_SHAPEINFERENCE=1 buck test caffe2/caffe2/python/operator_test:elementwise_ops_test
CAFFE2_ASSERT_SHAPEINFERENCE=1 buck test caffe2/caffe2/python/operator_test:math_ops_test

Note that the tests have to be executed with CAFFE2_ASSERT_SHAPEINFERENCE=1 in order to fail upon shape inference failure.

Differential Revision: D19880164

@facebook-github-bot
[Caffe2] Fix shape inference for element-wise operators
a22d714 
Summary:
Some elementwise operators don't have shape and type inference specified for the output tensor: `BitwiseOr`, `BitwiseAnd`, `BitwiseXor`, `Not`, `Sign`.

This change fixes this issue:
- For `Not` and `Sign` operators, the output has the same type and shape as the input, so `IdenticalTypeAndShapeOfInput` function is used to specify that.
- For bitwise operators created by `CAFFE2_SCHEMA_FOR_BINARY_BITWISE_OP` macro, the type and shape inference rules should be the same as for other binary element-wise operators, so `TensorInferenceFunction(ElementwiseOpShapeInference)` is used to specify that.

Also some tests were modified to ensure that the shape and type are inferred (`ensure_outputs_are_inferred` parameter)

Test Plan:
```
CAFFE2_ASSERT_SHAPEINFERENCE=1 buck test caffe2/caffe2/python/operator_test:elementwise_ops_test
CAFFE2_ASSERT_SHAPEINFERENCE=1 buck test caffe2/caffe2/python/operator_test:math_ops_test
```

Note that the tests have to be executed with `CAFFE2_ASSERT_SHAPEINFERENCE=1` in order to fail upon shape inference failure.

Differential Revision: D19880164

fbshipit-source-id: 5e64b60f2f2f072a315cd9d8281c467403616944
@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D19880164

@dr-ci
Copy link

dr-ci bot commented Feb 18, 2020

💊 CircleCI build failures summary and remediations

As of commit a22d714:

  • 1/1 failures introduced in this PR

Detailed failure analysis

One may explore the probable reasons each build failed interactively on the Dr. CI website.

🕵️ 1 new failure recognized by patterns

The following build failures do not appear to be due to upstream breakage:

See CircleCI build pytorch_windows_vs2019_py36_cuda10.1_test2 (1/1)

Step: "Test" (full log | pattern match details)

RuntimeError: test_jit_fuser failed! 
 
---------------------------------------------------------------------- 
Ran 46 tests in 11.727s 
 
FAILED (errors=4, skipped=10) 
Traceback (most recent call last): 
  File "run_test.py", line 486, in <module> 
    main() 
  File "run_test.py", line 479, in main 
    raise RuntimeError(message) 
RuntimeError: test_jit_fuser failed! 
 
(base) circleci@PACKER-5E29F737 C:\Users\circleci\project\test>if ERRORLEVEL 1 exit /b 1  
+ cleanup
+ retcode=1
+ 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.

@facebook-github-bot
Copy link
Contributor

This pull request has been merged in ee23944.

1 similar comment
@facebook-github-bot
Copy link
Contributor

This pull request has been merged in ee23944.

hczhu pushed a commit that referenced this pull request Feb 28, 2020
[Caffe2] Fix shape inference for element-wise operators (#33431)
e202ea8 
Summary:
Pull Request resolved: #33431

Some elementwise operators don't have shape and type inference specified for the output tensor: `BitwiseOr`, `BitwiseAnd`, `BitwiseXor`, `Not`, `Sign`.

This change fixes this issue:
- For `Not` and `Sign` operators, the output has the same type and shape as the input, so `IdenticalTypeAndShapeOfInput` function is used to specify that.
- For bitwise operators created by `CAFFE2_SCHEMA_FOR_BINARY_BITWISE_OP` macro, the type and shape inference rules should be the same as for other binary element-wise operators, so `TensorInferenceFunction(ElementwiseOpShapeInference)` is used to specify that.

Also some tests were modified to ensure that the shape and type are inferred (`ensure_outputs_are_inferred` parameter)

Test Plan:
```
CAFFE2_ASSERT_SHAPEINFERENCE=1 buck test caffe2/caffe2/python/operator_test:elementwise_ops_test
CAFFE2_ASSERT_SHAPEINFERENCE=1 buck test caffe2/caffe2/python/operator_test:math_ops_test
```

Note that the tests have to be executed with `CAFFE2_ASSERT_SHAPEINFERENCE=1` in order to fail upon shape inference failure.

Reviewed By: idning

Differential Revision: D19880164

fbshipit-source-id: 5d7902e045d79e5669e5e98dfb13a39711294939
ttumiel pushed a commit to ttumiel/pytorch that referenced this pull request Mar 4, 2020
@ttumiel
[Caffe2] Fix shape inference for element-wise operators (pytorch#33431)
24f3d8c 
Summary:
Pull Request resolved: pytorch#33431

Some elementwise operators don't have shape and type inference specified for the output tensor: `BitwiseOr`, `BitwiseAnd`, `BitwiseXor`, `Not`, `Sign`.

This change fixes this issue:
- For `Not` and `Sign` operators, the output has the same type and shape as the input, so `IdenticalTypeAndShapeOfInput` function is used to specify that.
- For bitwise operators created by `CAFFE2_SCHEMA_FOR_BINARY_BITWISE_OP` macro, the type and shape inference rules should be the same as for other binary element-wise operators, so `TensorInferenceFunction(ElementwiseOpShapeInference)` is used to specify that.

Also some tests were modified to ensure that the shape and type are inferred (`ensure_outputs_are_inferred` parameter)

Test Plan:
```
CAFFE2_ASSERT_SHAPEINFERENCE=1 buck test caffe2/caffe2/python/operator_test:elementwise_ops_test
CAFFE2_ASSERT_SHAPEINFERENCE=1 buck test caffe2/caffe2/python/operator_test:math_ops_test
```

Note that the tests have to be executed with `CAFFE2_ASSERT_SHAPEINFERENCE=1` in order to fail upon shape inference failure.

Reviewed By: idning

Differential Revision: D19880164

fbshipit-source-id: 5d7902e045d79e5669e5e98dfb13a39711294939
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
fb-exported Merged
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@cheparukhin @facebook-github-bot @mruberry