Update alowed python versions for `onnx-graphsurgeon`

[tadejsv]

onnx-graphsurgeon only works with python versions up to 3.8. Please upgrade it to at least 3.10 - this is the Python version used in all latest Nvidia containers

[lix19937]

pip install nvidia-pyindex
pip install onnx-graphsurgeon

[lix19937]

or https://developer.nvidia.com/tensorrt/download then unzip
This directory will have sub-directories like lib, include, data, and so on.

ls TensorRT-${version}   

bin  data  doc  include  lib  onnx_graphsurgeon  python  samples  targets

check the version of onnx_graphsurgeon

pip install onnx_graphsurgeon-0.5.1-py2.py3-none-any.whl

https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html#installing-tar

[tadejsv]

Thanks, I don't want to download and unzip the file manually, I want to do it using pip so that I can properly specify my environment and resolve dependencies. Please fix the issue with missing versions in NVIDIA's python package index

[zerollzeng]

Checking internally.

[zerollzeng]

onnx-graphsurgeon only works with python versions up to 3.8.

it should be python version agnostic, could you please provide a reproduce of the issue? thanks!

[tadejsv]

onnx-graphsurgeon only works with python versions up to 3.8.

it should be python version agnostic, could you please provide a reproduce of the issue? thanks!

Yeah you're right - this was actually a problem with newer versions of onnx-graphsurgeon being available. However, you can still change.

However, it would be good to update classifiers in setup.py to the python versions actually supported:

https://github.com/NVIDIA/TensorRT/blob/28733f0fdccde2967fed395b06ca491af3a561a9/tools/onnx-graphsurgeon/setup.py#L48-L55

[zerollzeng]

@pranavm-nvidia ^ ^