A Jupyter notebook testing DirectML support of ONNX on Windows using an Intel GPU.
Initialize the virtual environment and start Jupyter Lab. Open an Anaconda prompt (Miniconda is sufficient) and run:
conda env create -f conda.yaml
conda activate onnx-directml
python -m ipykernel install --user --name=onnx-directml
jupyter lab
Then open notebook.ipynb
and continue there.
Note: if Jupyter does not load on Chrome, open chrome://net-internals/#hsts and use
Delete domain security policies
to deletelocalhost
, then try again.