Error: `tf.gradients is not supported when eager execution is enabled`

[rstudio-conf-2020]

Error: `tf.gradients is not supported when eager execution is enabled`

[bradleyboehmke]

I was catching up on the CNN visualization notebooks today. There's two ways to resolve this:

1. Using TF 2.0 GradientTape as in here.
2. Disabling eager execution at the top of the script with tf$compat$v1$disable_eager_execution().

Neither of them have side effects on the rest of the notebook code chunks so I'll probably just use tf$compat$v1$disable_eager_execution().

[bradleyboehmke]

Good catch @OmaymaS. That is a notebook that I haven't really done anything with. I grabbed it from JJ Allaire's book repo but planned to revise it and base it on the Cats vs. Dogs model. When I do the revision I will try work out this issue.

[DiabbZegpi]

I was catching up on the CNN visualization notebooks today. There's two ways to resolve this:

1. Using TF 2.0 GradientTape as in here.
2. Disabling eager execution at the top of the script with tf$compat$v1$disable_eager_execution().

Neither of them have side effects on the rest of the notebook code chunks so I'll probably just use tf$compat$v1$disable_eager_execution().

Doing tihs also results in an error (I'm also following the cats vs dogs example on "Visualizing convnet filters"):
Error in py_call_impl():

! TypeError: You are passing KerasTensor(type_spec=TensorSpec(shape=(), dtype=tf.float32, name=None), name='tf.math.reduce_mean/Mean:0', description="created by layer 'tf.math.reduce_mean'"), an intermediate Keras symbolic input/output, to a TF API that does not allow registering custom dispatchers, such as `tf.cond`, `tf.function`, gradient tapes, or `tf.map_fn`. Keras Functional model construction only supports TF API calls that *do* support dispatching, such as `tf.math.add` or `tf.reshape`. Other APIs cannot be called directly on symbolic Kerasinputs/outputs. You can work around this limitation by putting the operation in a custom Keras layer `call` and calling that layer on this symbolic input/output.