-
Notifications
You must be signed in to change notification settings - Fork 109
Keras2onnx error with EfficientNet - _remove_unused_nodes and input_1:01 is disconnected #713
Comments
I am also facing an issue with the conversion of the EfficientNet model. It gives an error:
And when I tried the example script it gives me the following error:
|
Sorry to ask, but, have any mantainers seen this issue? |
Hi @jhelsas, I have tried a slightly different approach to convert Keras model to Onnx. To give an overview, I have done transfer learning on the Keras EfficientNet model. First, I convert my Keras model to Tensorflow Model using:
Then I use tf2onnx to convert my Tensorflow model to
When converting the EfficientModel is gives several gradient warning (reference here)
And when I tried to load the Onnx model for inference, it is predicting as expected using:
Hope that helped |
Dear keras2onnx mantainers,
I am trying to convert a keras model to onnx format using your library, but I am having trouble. I first tried with my model, then I tried just running the example notebook and it failed. the error on my model is the following:
Which is essentially the same as the one I found running the example notebook in here :
I have tried with tensorflow 2.4.0 and 2.1.0, the later recommended in the notebook.
I don't really know how to proceed from here, if you have any suggestion it would be greatly appreciated.
Best regards,
José hugo Elsas
The text was updated successfully, but these errors were encountered: