Loading the weight file with use_weight_norm parameter using the updated compiled_tcn function

[hlmhlr]

I used earlier version of this repo, trained the TCN model with following parameters, and installed libraries version are (tensorflow==2.11.1 and keras-tcn=3.5.0):

model = compiled_tcn(
        return_sequences=False,
         num_feat=4, # 
        num_classes=args.num_classes,  # from CLI
        nb_filters=40,
        kernel_size=8,
        dilations=[2 ** i for i in range(9)],
        nb_stacks=1,
        max_len=x_train[0:1].shape[1],
        use_weight_norm=True,
        use_skip_connections=True,
        dropout_rate=0.05,
        opt=args.optimizer, # from CLI
        lr=args.lr
    )

For some reasons, I had to shift the code onto the colab notebook, where I tried to install the same version of tensorflow and keras-tcn, but it only allowed me to install the tensorflow version greater than or equal to 2.12 and the updated version of keras-tcn.

Now, the latest repo of keras-tcn has removed the parameter use_weight_norm from compiled_tcn function, and therefore, I am unable to load the weights which I saved with the model mentioned above, and thus to evaluate the data.

How can I load the weight file (in which the use_weight_norm parameter was used) with the latest compiled_tcn mdel function from which the use_weight_norm parameter has been removed.