[torchlib] Fix layer norm dtype

[justinchuby]

Fix layer norm dtype mismatch errors

Fixes #2099

[Copilot]

Pull Request Overview

This PR addresses a layer normalization datatype mismatch error by updating how constants are created for weight initialization and removing an unnecessary helper function.

• Updated weight initialization to ensure that the constant matches the input tensor’s dtype.
• Removed the bias initialization block and the auxiliary _aten_layer_norm_onnx function, directly calling op.LayerNormalization.

Comments suppressed due to low confidence (1)

onnxscript/function_libs/torch_lib/ops/core.py:4755

• The bias initialization block was removed, so if bias is None, op.LayerNormalization will receive None. Please verify that passing None for bias is acceptable or provide a suitable default.

result, _, _ = op.LayerNormalization(input, weight, bias, axis=axis, epsilon=eps)

[codecov]

❌ 3 Tests Failed:

Tests completed	Failed	Passed	Skipped
11764	3	11761	2406

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

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0963_test_resize_downsample_scales_linear_antialias

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_resize_downsample_scales_linear_antialias'

The above exception was the direct cause of the following exception:
.nox\test_torch_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_resize_downsample_scales_linear_antialias' (e=No module named 'tests.onnx_backend_test_code.test_resize_downsample_scales_linear_antialias') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_resize_downsample_scales_linear_antialias.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_resize_downsample_scales_linear_antialias.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 opset19
E   
E   @script()
E   def bck_test_resize_downsample_scales_linear_antialias(X: FLOAT[1,1,4,4], scales: FLOAT[4]) -> (FLOAT[1,1,2,2]):
E       Y = opset19.Resize(X, None, scales, antialias=1, mode='linear')
E       return Y

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_1241_test_tfidfvectorizer_tf_batch_onlybigrams_skip5

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_tfidfvectorizer_tf_batch_onlybigrams_skip5'

The above exception was the direct cause of the following exception:
.nox\test_torch_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_tfidfvectorizer_tf_batch_onlybigrams_skip5' (e=No module named 'tests.onnx_backend_test_code.test_tfidfvectorizer_tf_batch_onlybigrams_skip5') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_tfidfvectorizer_tf_batch_onlybigrams_skip5.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_tfidfvectorizer_tf_batch_onlybigrams_skip5.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, INT32
E   from onnxscript.onnx_opset import opset9
E   
E   @script()
E   def bck_test_tfidfvectorizer_tf_batch_onlybigrams_skip5(X: INT32[2,6]) -> (FLOAT[2,7]):
E       Y = opset9.TfIdfVectorizer(X, max_gram_length=2, max_skip_count=5, min_gram_length=2, mode='TF', ngram_counts=[0, 4], ngram_indexes=[0, 1, 2, 3, 4, 5, 6], pool_int64s=[2, 3, 5, 4, 5, 6, 7, 8, 6, 7])
E       return Y

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0000_test_abs

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_abs'

The above exception was the direct cause of the following exception:
.nox\test_torch_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_abs' (e=No module named 'tests.onnx_backend_test_code.test_abs') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_abs.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_abs.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 opset13
E   
E   @script()
E   def bck_test_abs(x: FLOAT[3,4,5]) -> (FLOAT[3,4,5]):
E       y = opset13.Abs(x)
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.}