Unable to Convert CycleGAN Model into Intermediate Representation (IR)
Content Type: Troubleshooting | Article ID: 000092476 | Last Reviewed: 06/21/2023
The model conversion might have failed due to uninitialized variables.
Possible workarounds are as below:
Initialize all variables and reload the weights.
or
Freeze the graph and export it to protobuf (.pb) file. Refer to Freezing Custom Models in Python*.
Starting from 2021.3, TensorFlow* 1 is supported with limitations, refer to the Release Notes.