[torchlib] Fix calls to Unsqueeze to provide correct 1d axes

[justinchuby]

Discovered in onnx/onnx#6886 (comment), the axes input in calls to unsqueeze are sometimes 0d. This is incorrect according to the ONNX spec. The PR fixes the instances I could find.

[Copilot]

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (3)

onnxscript/function_libs/torch_lib/ops/special.py:222

• [nitpick] Consider using the 'axes' keyword for clarity (e.g., op.Unsqueeze(self, axes=[0])) to align with other parts of the codebase.

self = op.Unsqueeze(self, [0])

onnxscript/function_libs/torch_lib/ops/nn.py:1005

• [nitpick] For consistency, consider adopting the use of the 'axes' keyword in both Unsqueeze and Squeeze calls throughout the file.

self = op.Unsqueeze(self, [0])

onnxscript/function_libs/torch_lib/ops/core.py:7882

• [nitpick] To maintain a consistent API usage across the codebase, consider using the 'axes' keyword (e.g., op.Unsqueeze(index_base, axes=[-1])).

index_base = op.Unsqueeze(index_base, [-1])

[codecov]

❌ 4 Tests Failed:

Tests completed	Failed	Passed	Skipped
13927	4	13923	1882

View the top 3 failed test(s) by shortest run time

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0388_test_gather_1

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\3.11.9\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_gather_1'

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_gather_1' (e=No module named 'tests.onnx_backend_test_code.test_gather_1') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_gather_1.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_gather_1.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_1(data: FLOAT[5,4,3,2], indices: INT64[3]) -> (FLOAT[5,3,3,2]):
E       y = opset13.Gather(data, indices, axis=1)
E       return y

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0507_test_layer_normalization_3d_axis0_epsilon

Stack Traces | 0.005s 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_layer_normalization_3d_axis0_epsilon'

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_layer_normalization_3d_axis0_epsilon' (e=No module named 'tests.onnx_backend_test_code.test_layer_normalization_3d_axis0_epsilon') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_layer_normalization_3d_axis0_epsilon.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_layer_normalization_3d_axis0_epsilon.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 opset17
E   
E   @script()
E   def bck_test_layer_normalization_3d_axis0_epsilon(X: FLOAT[2,3,5], W: FLOAT[2,3,5], B: FLOAT[2,3,5]) -> (FLOAT[2,3,5], FLOAT[1,1,1], FLOAT[1,1,1]):
E       Y, Mean, InvStdDev = opset17.LayerNormalization(X, W, B, axis=0, epsilon=0.10000000149011612)
E       return Y, Mean, InvStdDev

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0869_test_reduce_max_do_not_keepdims_random

Stack Traces | 0.006s run time

onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.10.11\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_reduce_max_do_not_keepdims_random'

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_reduce_max_do_not_keepdims_random' (e=No module named 'tests.onnx_backend_test_code.test_reduce_max_do_not_keepdims_random') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_reduce_max_do_not_keepdims_random.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_reduce_max_do_not_keepdims_random.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 opset18
E   
E   @script()
E   def bck_test_reduce_max_do_not_keepdims_random(data: FLOAT[3,2,2], axes: INT64[1]) -> (FLOAT[3,2]):
E       reduced = opset18.ReduceMax(data, axes, keepdims=0)
E       return reduced

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

[justinchuby]

It is now correct

[gramalingam]

I remember noticing something like this while writing some fusion patterns. May need to double-check fusion patterns in case they are impacted by this. (Just noting it down here before I forget!)