onnxmltools
onnxmltools copied to clipboard
Remove use of auto_pad in convert_keras and convert_coreml
Given that auto_pad is deprecated in ONNX and isn't supported for things like infer_shapes, I propose that we remove auto_pad from coreml and keras operator converters (Pool/Conv) and instead use explicit padding.
I was thinking about implementing it. Does anyone have any issues with this?
Since the keras has output shape, it's doable to calculate the actual padding size. And there is one reminder that onnxmltools needs to generate multiple version onnx model, e.g. any change must make sure the backward compatibility.
@pjohnson-wave I'm quite interested in this matter. Would you like to tell me how to modify the code?