Fix _to_model_proto after protobuf update

[xadupre]

No description provided.

[codecov]

❌ 2 Tests Failed:

Tests completed	Failed	Passed	Skipped
13681	2	13679	1417

View the full list of 2 ❄️ flaky test(s)

onnxscript.tools.transformers_models.phi_test.TestExportPhi::test_phi_export_cpu

Flake rate in main: 97.22% (Passed 19 times, Failed 665 times)

Stack Traces | 15.2s run time

..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:250: in call
    pass_result = pass_(model)
                  ^^^^^^^^^^^^
..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:127: in __call__
    result = self.call(model)
             ^^^^^^^^^^^^^^^^
onnxscript/version_converter/__init__.py:77: in call
    raise ValueError(
E   ValueError: The model contains functions. The version conversion pass does not support functions. Please use `common_passes.InlinePass` to inline the functions before applying this pass (_ConvertVersionPassRequiresInline).

The above exception was the direct cause of the following exception:
onnxscript/_internal/version_utils.py:94: in call_f
    return fct(self)
           ^^^^^^^^^
.../tools/transformers_models/phi_test.py:30: in test_phi_export_cpu
    proto = onnxscript.tools.transformers_models.export_to_onnx(model, *input_tensors)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
.../tools/transformers_models/__init__.py:41: in export_to_onnx
    prog = torch.onnx.export(model, args, dynamo=True)  # pylint: disable=no-value-for-parameter
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
..../test_onnx_ir_git/lib/python3.11.../torch/onnx/__init__.py:364: in export
    return _compat.export_compat(
..../test_onnx_ir_git/lib/python3.11.../_internal/exporter/_compat.py:180: in export_compat
    onnx_program.model = onnxscript_apis.convert_version(
onnxscript/_framework_apis/torch_2_6.py:43: in convert_version
    version_converter.convert_version(model, target_version, fallback=True)
onnxscript/version_converter/__init__.py:173: in convert_version
    ConvertVersionPass(target_version=target_version, fallback=fallback)(model)
..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:127: in __call__
    result = self.call(model)
             ^^^^^^^^^^^^^^^^
onnxscript/version_converter/__init__.py:52: in call
    return self.convert_pass(model)
           ^^^^^^^^^^^^^^^^^^^^^^^^
..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:127: in __call__
    result = self.call(model)
             ^^^^^^^^^^^^^^^^
..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:253: in call
    raise PassError(
E   onnx_ir.passes.PassError: An error occurred when running the '<onnxscript.version_converter._ConvertVersionPassRequiresInline object at 0x1453cecd0>' pass after the following passes: ['<onnx_ir.passes.common.inliner.InlinePass object at 0x145339e10>']

onnxscript.tools.transformers_models.phi_test.TestExportPhi::test_phi_export_cpu_export_api

Flake rate in main: 97.22% (Passed 19 times, Failed 665 times)

Stack Traces | 5.2s run time

..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:250: in call
    pass_result = pass_(model)
                  ^^^^^^^^^^^^
..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:127: in __call__
    result = self.call(model)
             ^^^^^^^^^^^^^^^^
onnxscript/version_converter/__init__.py:77: in call
    raise ValueError(
E   ValueError: The model contains functions. The version conversion pass does not support functions. Please use `common_passes.InlinePass` to inline the functions before applying this pass (_ConvertVersionPassRequiresInline).

The above exception was the direct cause of the following exception:
onnxscript/_internal/version_utils.py:94: in call_f
    return fct(self)
           ^^^^^^^^^
.../tools/transformers_models/phi_test.py:48: in test_phi_export_cpu_export_api
    proto = onnxscript.tools.transformers_models.export_to_onnx(
.../tools/transformers_models/__init__.py:41: in export_to_onnx
    prog = torch.onnx.export(model, args, dynamo=True)  # pylint: disable=no-value-for-parameter
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
..../test_onnx_ir_git/lib/python3.11.../torch/onnx/__init__.py:364: in export
    return _compat.export_compat(
..../test_onnx_ir_git/lib/python3.11.../_internal/exporter/_compat.py:180: in export_compat
    onnx_program.model = onnxscript_apis.convert_version(
onnxscript/_framework_apis/torch_2_6.py:43: in convert_version
    version_converter.convert_version(model, target_version, fallback=True)
onnxscript/version_converter/__init__.py:173: in convert_version
    ConvertVersionPass(target_version=target_version, fallback=fallback)(model)
..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:127: in __call__
    result = self.call(model)
             ^^^^^^^^^^^^^^^^
onnxscript/version_converter/__init__.py:52: in call
    return self.convert_pass(model)
           ^^^^^^^^^^^^^^^^^^^^^^^^
..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:127: in __call__
    result = self.call(model)
             ^^^^^^^^^^^^^^^^
..../test_onnx_ir_git/lib/python3.11.../onnx_ir/passes/_pass_infra.py:253: in call
    raise PassError(
E   onnx_ir.passes.PassError: An error occurred when running the '<onnxscript.version_converter._ConvertVersionPassRequiresInline object at 0x1455c0710>' pass after the following passes: ['<onnx_ir.passes.common.inliner.InlinePass object at 0x1455cfe50>']

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]

The functions argument was removed. Was it used anywhere?

[xadupre]

You can look into the unittest. def to_model_proto(self, **kwargs):, the signature is not very explicit about what is allowed and what is not.

[justinchuby]

There isn’t a test failing? I thought I removed them all. We can update the signature to reflect the api change. (It was mentioned in the release notes)