Skip to content

[torchlib] Fix pow.Tensor_Scalar type promotion #2335

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

Merged
merged 5 commits into from
May 28, 2025
Merged

Conversation

justinchuby
Copy link
Collaborator

@justinchuby justinchuby commented May 23, 2025

Fix pow.Tensor_Scalar type promotion by accounting different combination of input dtypes. This change ensures the inputs to Pow is always the same type for compatibility with downstream tools.

Also

Fix #2213

Copy link
Contributor

@Copilot Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull Request Overview

This PR fixes type promotion in the pow.Tensor_Scalar operation by ensuring that both inputs to the Pow operator have matching dtypes. In addition, it updates related tests and improves IR string representation and enum capabilities.

  • Updated tests in torch_lib to cover various tensor-scalar pow operations.
  • Added a new is_floating_point method to the data type enum.
  • Modified the aten_pow_tensor_scalar implementation to handle type promotion for both floating point and integer cases.

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File Description
tests/function_libs/torch_lib/e2e_ops_tests.py Updated test cases to assert ONNX program outputs for various pow scenarios.
onnxscript/ir/_enums.py Added a new is_floating_point method for better data type handling.
onnxscript/ir/_core.py Refactored the usage of _short_tensor_str for cleaner IR string representation.
onnxscript/function_libs/torch_lib/ops/core.py Revised aten_pow_tensor_scalar to include explicit type promotion logic.

@justinchuby justinchuby added module: torchlib Related to the torch/aten function lib in development module: IR Intermediate representation labels May 23, 2025
@justinchuby justinchuby mentioned this pull request May 23, 2025
9 tasks
Copy link

codecov bot commented May 23, 2025

❌ 4 Tests Failed:

Tests completed Failed Passed Skipped
16936 4 16932 2345
View the top 3 failed test(s) by shortest run time
onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0175_test_cast_FLOAT16_to_DOUBLE
Stack Traces | 0.004s run time
onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.12.10\x64\Lib\importlib\__init__.py:90: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
E   ModuleNotFoundError: No module named 'tests.onnx_backend_test_code.test_cast_FLOAT16_to_DOUBLE'

The above exception was the direct cause of the following exception:
.nox\test\Lib\site-packages\parameterized\parameterized.py:620: in standalone_func
    return func(*(a + p.args), **p.kwargs, **kw)
onnxscript\backend\onnx_export_test.py:271: in test_export2python_produces_correct_onnx_script_model
    functions = extract_functions(backend_test.name, code, self.test_folder)
onnxscript\backend\onnx_export_test.py:139: in extract_functions
    raise AssertionError(
E   AssertionError: Unable to import 'tests.onnx_backend_test_code.test_cast_FLOAT16_to_DOUBLE' (e=No module named 'tests.onnx_backend_test_code.test_cast_FLOAT16_to_DOUBLE') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_cast_FLOAT16_to_DOUBLE.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_cast_FLOAT16_to_DOUBLE.py', current folder: D:\a\onnxscript\onnxscript
E   ---- CONTENT --
E   import numpy
E   from onnx import TensorProto
E   from onnx.helper import make_tensor
E   from onnxscript import script, external_tensor
E   from onnxscript.values import Opset
E   from onnxscript.onnx_types import DOUBLE, FLOAT16
E   from onnxscript.onnx_opset import opset21
E   
E   @script()
E   def bck_test_cast_FLOAT16_to_DOUBLE(input: FLOAT16[3,4]) -> (DOUBLE[3,4]):
E       output = opset21.Cast(input, to=11)
E       return output
onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0393_test_gather_negative_indices
Stack Traces | 0.004s run time
onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.12.10\x64\Lib\importlib\__init__.py:90: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
E   ModuleNotFoundError: No module named 'tests.onnx_backend_test_code.test_gather_negative_indices'

The above exception was the direct cause of the following exception:
.nox\test\Lib\site-packages\parameterized\parameterized.py:620: in standalone_func
    return func(*(a + p.args), **p.kwargs, **kw)
onnxscript\backend\onnx_export_test.py:271: in test_export2python_produces_correct_onnx_script_model
    functions = extract_functions(backend_test.name, code, self.test_folder)
onnxscript\backend\onnx_export_test.py:139: in extract_functions
    raise AssertionError(
E   AssertionError: Unable to import 'tests.onnx_backend_test_code.test_gather_negative_indices' (e=No module named 'tests.onnx_backend_test_code.test_gather_negative_indices') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_gather_negative_indices.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_gather_negative_indices.py', current folder: D:\a\onnxscript\onnxscript
E   ---- CONTENT --
E   import numpy
E   from onnx import TensorProto
E   from onnx.helper import make_tensor
E   from onnxscript import script, external_tensor
E   from onnxscript.values import Opset
E   from onnxscript.onnx_types import FLOAT, INT64
E   from onnxscript.onnx_opset import opset13
E   
E   @script()
E   def bck_test_gather_negative_indices(data: FLOAT[10], indices: INT64[3]) -> (FLOAT[3]):
E       y = opset13.Gather(data, indices, axis=0)
E       return y
onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0772_test_prelu_broadcast
Stack Traces | 0.004s run time
onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.12.10\x64\Lib\importlib\__init__.py:90: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
E   ModuleNotFoundError: No module named 'tests.onnx_backend_test_code.test_prelu_broadcast'

The above exception was the direct cause of the following exception:
.nox\test\Lib\site-packages\parameterized\parameterized.py:620: in standalone_func
    return func(*(a + p.args), **p.kwargs, **kw)
onnxscript\backend\onnx_export_test.py:271: in test_export2python_produces_correct_onnx_script_model
    functions = extract_functions(backend_test.name, code, self.test_folder)
onnxscript\backend\onnx_export_test.py:139: in extract_functions
    raise AssertionError(
E   AssertionError: Unable to import 'tests.onnx_backend_test_code.test_prelu_broadcast' (e=No module named 'tests.onnx_backend_test_code.test_prelu_broadcast') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_prelu_broadcast.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_prelu_broadcast.py', current folder: D:\a\onnxscript\onnxscript
E   ---- CONTENT --
E   import numpy
E   from onnx import TensorProto
E   from onnx.helper import make_tensor
E   from onnxscript import script, external_tensor
E   from onnxscript.values import Opset
E   from onnxscript.onnx_types import FLOAT
E   from onnxscript.onnx_opset import opset16
E   
E   @script()
E   def bck_test_prelu_broadcast(x: FLOAT[3,4,5], slope: FLOAT[5]) -> (FLOAT[3,4,5]):
E       y = opset16.PRelu(x, slope)
E       return y

To view more test analytics, go to the Test Analytics Dashboard
📋 Got 3 mins? Take this short survey to help us improve Test Analytics.

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature.
@justinchuby justinchuby modified the milestones: 0.2.7, 0.3 May 23, 2025
@justinchuby justinchuby marked this pull request as draft May 27, 2025 20:35
@justinchuby justinchuby marked this pull request as ready for review May 27, 2025 20:35

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature.
@justinchuby justinchuby removed this from the 0.2.7 milestone May 28, 2025
@justinchuby justinchuby added this to the 0.3 milestone May 28, 2025
justinchuby added a commit to onnx/ir-py that referenced this pull request May 28, 2025

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature.
Migration of changes in microsoft/onnxscript#2335

Signed-off-by: Justin Chu <justinchuby@users.noreply.github.com>
Copy link
Contributor

@titaiwangms titaiwangms left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Have we already deprecated type promotion in exporter? I don't remember the status of type promotion.

justinchuby added a commit to onnx/ir-py that referenced this pull request May 28, 2025

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature.
Migration of changes in
microsoft/onnxscript#2335

Signed-off-by: Justin Chu <justinchuby@users.noreply.github.com>
@justinchuby justinchuby enabled auto-merge (squash) May 28, 2025 21:37
@justinchuby justinchuby merged commit 61d4ab5 into main May 28, 2025
24 of 29 checks passed
@justinchuby justinchuby deleted the justinchu/fix-pow branch May 28, 2025 21:37
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
module: IR Intermediate representation module: torchlib Related to the torch/aten function lib in development
Projects
Development

Successfully merging this pull request may close these issues.

[torchlib] Pow(int, float) isn't converted correctly
2 participants