Xavier Dupré
Xavier Dupré
It looks good to me. cc @gramalingam
There is. It consists in storing the weights outside the onnx graph, see https://onnx.ai/onnx/repo-docs/ExternalData.html.
Why `reg_schema` and not `register_schema` which seems easier to understand? Should we check that `has_schema` returns True as well? Should we add a method to unregister a schema?
What is FINN-ONNX? You should be able to use onnx python api to modify your onnx file: https://onnx.ai/onnx/intro/python.html.
Is it possible to express as a function of [DFT](https://onnx.ai/onnx/operators/onnx__DFT.html)?
It would be a nice feature to have.
Is it on CPU or GPU? This PR addresses one issue on GPU: https://github.com/microsoft/onnxruntime/pull/19540.
Maybe it is worth adding a unit test.
mlprodict is archived because its main functionality, a python runtime for onnx called OnnxInference, was moved into onnx package as ReferenceEvaluator. I did not write an example about the trick...
Every model used in unit tests usuaaly specifies the ir_version to avoid onnxruntime to fail because it does not support the new one yet. Any model already used in the...