Cuda Toolkit Required?

[rallen911]

I have an existing program that creates a keras model via: model = tf.keras.Sequential(layer) The model was being saved as a .h5, so I added: model.save("model_currentbest"+".tf") to save it in a supported format.

When I run the conversion from the command line using: python -m tf2onnx.convert --saved-model model_currentbest.tf --output model.onnx , I get the output in the attached file. The last part of the trace contains: File "C:\Users\Rob\AppData\Roaming\Python\Python38\site-packages\tensorflow\python\saved_model\load.py", line 448, in _load_nodes slot_variable=optimizer_object.add_slot( AttributeError: '_UserObject' object has no attribute 'add_slot'

I'm wondering if I need to have the cuda toolkit installed too?

Regards,

Robert Allen
output.txt