incompatible architecture v1.14.0

[babenek]

Describe the issue

Specific configuration (latest) fails with the version.
Python 3.9.16 pip 23 onnxruntime 1.14.0

https://github.com/babenek/CredSweeper/actions/runs/4149456697/jobs/7178418969

ImportError:
dlopen(/Users/runner/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/onnxruntime/capi/onnxruntime_pybind11_state.so, 0x0002): tried: '/Users/runner/hostedtoolcache/Python/3.9.16/x64/lib/python3.9/site-packages/onnxruntime/capi/onnxruntime_pybind11_state.so' 
(mach-o file, but is an incompatible architecture (have (arm64), need (x86_64)))

To reproduce

GitHub workflow for macos with Python 3.9.16 and import onnxruntime 1.14.0

Urgency

Not urgent. Prrevious version will be used.

Platform

Mac

OS Version

macos 12.6.3 21G419

ONNX Runtime Installation

Released Package

ONNX Runtime Version or Commit ID

v1.14.0

ONNX Runtime API

Python

Architecture

X64

Execution Provider

Other / Unknown

Execution Provider Library Version

No response

[davidmezzetti]

Same. You know what they say about Friday releases.

https://github.com/neuml/txtai/actions/runs/4148508839/jobs/7178595222

[fxmarty]

We hit the same issue, see for example: https://github.com/huggingface/optimum/actions/runs/4163709490/jobs/7204436534

[jcwchen]

ONNX pipeline hits the same issue while using ORT 1.14 Python 3.9 wheel on Mac. See the log here: https://github.com/onnx/onnx/actions/runs/4147821575/jobs/7208166327.

[faxu]

Thanks for posting the issue. We are investigating..

[jchen351]

Updated python packages will be released along with v-1.14.1

[jcwchen]

I can confirm the fresh ORT patch 1.14.1 doesn't have such an issue anymore in ONNX pipelines: onnx/onnx#4950. Thanks @jchen351 for fixing this.

[babenek]

Confirm, The issue was fixed.
Thanks @jchen351, @davidmezzetti, @faxu, @fxmarty, @jcwchen for the support!

[varunsha30]

Hello, I am facing the same issue. I have installed onnxruntime using pip to my virtual environment. When I import onnxruntime using import onnxruntime it throws the following error:

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
Cell In[31], line 1
----> 1 import onnxruntime

File [~/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/__init__.py:55](https://file+.vscode-resource.vscode-cdn.net/Users/varunsharma/Downloads/Projects/LLM/~/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/__init__.py:55)
     52 from onnxruntime.capi import onnxruntime_validation
     54 if import_capi_exception:
---> 55     raise import_capi_exception
     57 from onnxruntime.capi.onnxruntime_inference_collection import InferenceSession  # noqa: F401
     58 from onnxruntime.capi.onnxruntime_inference_collection import IOBinding  # noqa: F401

File [~/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/__init__.py:23](https://file+.vscode-resource.vscode-cdn.net/Users/varunsharma/Downloads/Projects/LLM/~/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/__init__.py:23)
     13 # we need to do device version validation (for example to check Cuda version for an onnxruntime-training package).
     14 # in order to know whether the onnxruntime package is for training it needs
     15 # to do import onnxruntime.training.ortmodule first.
   (...)
     20 # meaningful messages to the user.
     21 # the saved exception is raised after device version validation.
     22 try:
---> 23     from onnxruntime.capi._pybind_state import ExecutionMode  # noqa: F401
     24     from onnxruntime.capi._pybind_state import ExecutionOrder  # noqa: F401
     25     from onnxruntime.capi._pybind_state import GraphOptimizationLevel  # noqa: F401

File [~/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/capi/_pybind_state.py:33](https://file+.vscode-resource.vscode-cdn.net/Users/varunsharma/Downloads/Projects/LLM/~/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/capi/_pybind_state.py:33)
     25         if not os.path.isfile(os.path.join(system_root, "System32", "vcruntime140_1.dll")):
     26             warnings.warn("Please install the 2019 Visual C++ runtime and then try again. "
     27                           "If you've installed the runtime in a non-standard location "
     28                           "(other than %SystemRoot%\System32), "
     29                           "make sure it can be found by setting the correct path.")
---> 33 from .onnxruntime_pybind11_state import *  # noqa

ImportError: dlopen(/Users/varunsharma/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/capi/onnxruntime_pybind11_state.so, 0x0002): tried: '/Users/varunsharma/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/capi/onnxruntime_pybind11_state.so' (mach-o file, but is an incompatible architecture (have 'arm64', need 'x86_64')), '/System/Volumes/Preboot/Cryptexes/OS/Users/varunsharma/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/capi/onnxruntime_pybind11_state.so' (no such file), '/Users/varunsharma/Downloads/Projects/LLM/llm/lib/python3.11/site-packages/onnxruntime/capi/onnxruntime_pybind11_state.so' (mach-o file, but is an incompatible architecture (have 'arm64', need 'x86_64'))

I am using m2 air with macos 14.0 (Sonoma beta version).
Here is some information about the onnxruntime installed:

Name: onnxruntime
Version: 1.15.1
Summary: ONNX Runtime is a runtime accelerator for Machine Learning models
Home-page: https://onnxruntime.ai
Author: Microsoft Corporation
Author-email: onnxruntime@microsoft.com
License: MIT License
Location: /Users/varunsharma/Downloads/Projects/LLM/llm/lib/python3.11/site-packages
Requires: coloredlogs, flatbuffers, numpy, packaging, protobuf, sympy
Required-by: chromadb