[torchlib] Fix various implementations

[justinchuby]

Fix implementations according to pytorch/pytorch#146224. Removed Eager Mode tests since we only care about the graph constructed. Changed all traceable ops to trace_only. Removed usage of the IsScalar function.

[codecov]

❌ 60 Tests Failed:

Tests completed	Failed	Passed	Skipped
10952	60	10892	1861

View the top 2 failed tests by shortest run time

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0307_test_cumsum_2d_axis_0

Stack Traces | 0.003s run time

onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.11.9\x64\Lib\importlib\__init__.py:126: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
E   ModuleNotFoundError: No module named 'tests.onnx_backend_test_code.test_cumsum_2d_axis_0'

The above exception was the direct cause of the following exception:
.nox\test_ort_nightly\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_cumsum_2d_axis_0' (e=No module named 'tests.onnx_backend_test_code.test_cumsum_2d_axis_0') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_cumsum_2d_axis_0.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_cumsum_2d_axis_0.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, INT32
E   from onnxscript.onnx_opset import opset14
E   
E   @script()
E   def bck_test_cumsum_2d_axis_0(x: DOUBLE[2,3], axis: INT32) -> (DOUBLE[2,3]):
E       y = opset14.CumSum(x, axis)
E       return y

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0758_test_or_bcast4v3d

Stack Traces | 0.003s run time

onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.11.9\x64\Lib\importlib\__init__.py:126: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
E   ModuleNotFoundError: No module named 'tests.onnx_backend_test_code.test_or_bcast4v3d'

The above exception was the direct cause of the following exception:
.nox\test_ort_nightly\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_or_bcast4v3d' (e=No module named 'tests.onnx_backend_test_code.test_or_bcast4v3d') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_or_bcast4v3d.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_or_bcast4v3d.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 BOOL
E   from onnxscript.onnx_opset import opset7
E   
E   @script()
E   def bck_test_or_bcast4v3d(x: BOOL[3,4,5,6], y: BOOL[4,5,6]) -> (BOOL[3,4,5,6]):
E       r_or = opset7.Or(x, y)
E       return r_or

View the full list of 1 ❄️ flaky tests

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_1286_test_xor_bcast4v3d

Flake rate in main: 6.12% (Passed 46 times, Failed 3 times)

Stack Traces | 0.003s run time

onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.11.9\x64\Lib\importlib\__init__.py:126: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
E   ModuleNotFoundError: No module named 'tests.onnx_backend_test_code.test_xor_bcast4v3d'

The above exception was the direct cause of the following exception:
.nox\test_ort_nightly\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_xor_bcast4v3d' (e=No module named 'tests.onnx_backend_test_code.test_xor_bcast4v3d') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_xor_bcast4v3d.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_xor_bcast4v3d.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 BOOL
E   from onnxscript.onnx_opset import opset7
E   
E   @script()
E   def bck_test_xor_bcast4v3d(x: BOOL[3,4,5,6], y: BOOL[4,5,6]) -> (BOOL[3,4,5,6]):
E       xor = opset7.Xor(x, y)
E       return xor

To view more test analytics, go to the Test Analytics Dashboard
📢 Thoughts on this report? Let us know!

[rrc-sony]

This breaks exporting scripts that use clamp(min=1e-12) (or any other float number) instead of a tensor, since floating point numbers don't have a .shape.
I expect the same issue appears in clamp_max but haven't tried to reproduce.

[justinchuby]

Thanks for reporting. I will create a patch today

[justinchuby]

#2066