Creating An ONNX Model
Introduction
ONNX GraphSurgeon includes an IR (intermediate representation) that can be exported to ONNX. Additionally, the IR exposes a simple API so that it is possible to build graphs manually.
This example creates an ONNX model containing a single GlobalLpPool node.
Running the example
Generate the model and save it to test_globallppool.onnx by running:
python3 example.py
The generated model will look like this:
