0 trainable parameters

[julietaumpierrez]

Issue Type

Others

OS

Linux

onnx2tf version number

1.18.14

onnx version number

1.14.1

onnxruntime version number

1.16.0

onnxsim (onnx_simplifier) version number

0.4.33

tensorflow version number

2.14.0

Download URL for ONNX

https://drive.google.com/file/d/1JCOpqBAV83Usxl-uaCSOzGEkwrjQxaA3/view?usp=share_link

Parameter Replacement JSON

-

Description

Hello,

I ran this command in linux: onnx2tf -i cnn42onnx.onnx -okv3 and then this code in python
model = tf.keras.models.load_model("saved_model/cnn42onnx_float32_v3.keras")

However when doing a model.summary() i get no trainable parameters. I want to be able to fine tune this model not only do inference, is there a way of obtaining a trainable model with your library?

Thanks again

[Katsuya Hyodo]

BatchNormalization and Dropout are lost in the process of model optimization; you can transcribe the weights yourself to the model structure defined in Keras.