AttributeError: 'Conv2D' object has no attribute 'apply'

[JosephMPedersen]

I'm not sure if I'm doing something wrong, but I just ran: pip install tensorflow-privacy
then: python mnist_dpsgd_tutorial.py

And I'm getting the error:

File "/privacy/tutorials/mnist_dpsgd_tutorial_common.py", line 23, in get_cnn_model

y = tf.keras.layers.Conv2D(
16, 8, strides=2, padding='same', activation='relu').apply(input_layer)

AttributeError: 'Conv2D' object has no attribute 'apply'

Here are the versions of the modules that I have installed:

tensorboard 2.9.0 pypi_0 pypi
tensorboard-data-server 0.6.1 pypi_0 pypi
tensorboard-plugin-wit 1.6.0 py_0
tensorflow 2.9.1 pypi_0 pypi
tensorflow-datasets 4.5.2 pypi_0 pypi
tensorflow-estimator 2.9.0 pypi_0 pypi
tensorflow-gpu 2.4.1 h30adc30_0
tensorflow-io-gcs-filesystem 0.26.0 pypi_0 pypi
tensorflow-metadata 1.8.0 pypi_0 pypi
tensorflow-privacy 0.8.0 pypi_0 pypi
tensorflow-probability 0.15.0 pypi_0 pypi

[JosephMPedersen]

Ok, I found that remove all of the ".apply" fixes the code. I'm not sure if I'm using an incompatible version of tensorflow, maybe Conv2D( ) use to return something with an apply( ) method, but for dependencies the page just states "installing TensorFlow (>= 1.14) is a pre-requisite", and I'm using 2.9.0

[peter-martin-17]

what does your code look like once you remove ".apply"?