Closed seidnerj closed 1 month ago
Thank you for this suggestion, but I think it would be better to do it in the following way:
from pathlib import Path
onnx_model_path = str(Path(self.saved_model_path).with_suffix('.onnx'))
Sure thing, I opened a new PR.
Currently Model2onnx expects a file with an extension of "h5" models (HDF5 format), but really there's not reason why this would be dependant on the file extension. This makes Model2onnx agnostic to the file extension (and also the tensorflow actual model type)