[Build] moduleNotfoundError: no module named 'onnxruntime.training' & 'No matching distribution found for onnxruntime-training'

[rajkamal-007]

Describe the issue

While I was trying to deploy the artifacts and run the onnx model on the edge device (Linux OS aarch64) its showing up with the following error:

I've followed the instructions for building the onnx runtime from the following link: https://onnxruntime.ai/docs/build/training.html#build-for-on-device-training
How to get rid of that error?

I even tried with the commands 'pip install onnxruntime-training'; 'pip3 install onnxruntime-training'; 'pip3.9 install onnx runtime-training' Nothing seems to work!

Urgency

Critical

Target platform

Linux (Am62x)

Build script

git clone --recursive https://github.com/Microsoft/onnxruntime.git
cd onnxruntime

./build.sh --config RelWithDebInfo --build_shared_lib --parallel --enable_training_apis

./build.sh --config RelWithDebInfo --build_shared_lib --parallel --enable_training

./build.sh --enable_training --use_dnnl

Error / output

Visual Studio Version

No response

GCC / Compiler Version

GCC 9.2.1

[carzh]

Ah, the commands pip install onnxruntime-training attempts to find the ONNXRuntime Training wheel from PyPi. We no longer upload ONNXRuntime Training packages to PyPi. If you would like to install the stable or nightly package, you can follow the installation directions from here (select Optimize Training -> On-device training -> Linux)

If you've followed the instructions for build, then you've also built a local version of the wheel file. To find it, go to your onnxruntime repository, then go to build/Linux/RelWithDebugInfo/dist and there should be an ONNXRuntime Training wheel file. To install it, run pip install [name of wheel file]

[snnn]

We only publish packages for x86_64. Sorry you need to build it from source.

[rajkamal-007]

I have tried to build from source also followed the commands for optimize training, but nothing seems to work.

[snnn]

As the error says, please upgrade your cmake. You can get the latest one from cmake.org

[rajkamal-007]

I've upgraded the cmake to 3.29.5, and tried to build using:
++++++++++++++++++++++++
git clone --recursive https://github.com/Microsoft/onnxruntime.git
cd onnxruntime
./build.sh --config RelWithDebInfo --build_shared_lib --parallel --enable_training --build_wheel --allow_running_as_root
++++++++++++++++++++++++++
Now its giving me the following error:

[snnn]

Could you show us more logs please?

[rajkamal-007]

Sure

[github-actions]

This issue has been automatically marked as stale due to inactivity and will be closed in 30 days if no further activity occurs. If further support is needed, please provide an update and/or more details.

[Leo5050xvjf]

Describe the issue

While I was trying to deploy the artifacts and run the onnx model on the edge device (Linux OS aarch64) its showing up with the following error:

I've followed the instructions for building the onnx runtime from the following link: https://onnxruntime.ai/docs/build/training.html#build-for-on-device-training How to get rid of that error?

I even tried with the commands 'pip install onnxruntime-training'; 'pip3 install onnxruntime-training'; 'pip3.9 install onnx runtime-training' Nothing seems to work!

Urgency

Critical

Target platform

Linux (Am62x)

Build script

git clone --recursive https://github.com/Microsoft/onnxruntime.git cd onnxruntime

./build.sh --config RelWithDebInfo --build_shared_lib --parallel --enable_training_apis

./build.sh --config RelWithDebInfo --build_shared_lib --parallel --enable_training

./build.sh --enable_training --use_dnnl

Error / output

Visual Studio Version

No response

GCC / Compiler Version

GCC 9.2.1

Hi, I encountered a similar problem. Have you solved this issue? Thank you!