Visualizing gradient for Keras model in R

I am trying to visualize gradient (gradcam) for a keras model in R.

Here's the code with MNIST dataset:

library(keras)
fashion_mnist <- dataset_fashion_mnist()

c(train_images, train_labels) %<-% fashion_mnist$train
c(test_images, test_labels) %<-% fashion_mnist$test
class_names = c('T-shirt/top',
                'Trouser',
                'Pullover',
                'Dress',
                'Coat',
                'Sandal',
                'Shirt',
                'Sneaker',
                'Bag',
                'Ankle boot')

train_images <- train_images / 255
test_images <- test_images / 255

model <- keras_model_sequential()
model %>%
  layer_flatten(input_shape = c(28, 28)) %>%
  layer_dense(units = 128, activation = 'relu') %>%
  layer_dense(units = 10, activation = 'softmax')

model %>% compile(
  optimizer = 'adam',
  loss = 'sparse_categorical_crossentropy',
  metrics = c('accuracy')
)

model %>% fit(train_images, train_labels, epochs = 5, verbose = 2)


img <- test_images[1, , , drop = FALSE]

predictions <- model %>% predict(img)


img_output <- model$output[, 1]

last_layer <- model %>% get_layer("dense_4")

grads <- k_gradients(img_output, last_layer$output)[[1]]

I am following these instructions: https://rstudio-conf-2020.github.io/dl-keras-tf/notebooks/visualizing-what-cnns-learn.nb.html

And I am getting this error:

RuntimeError: tf.gradients is not supported when eager execution is enabled. Use tf.GradientTape instead.

I found this solution from here:

tf$compat$v1$disable_eager_execution()

However, adding this to the top of the code produce the other error:

  AttributeError: 'Tensor' object has no attribute 'gradient'

while running this:

pooled_grads <- k_mean(grads$gradient, axis = c(1, 2, 3))

Can you help me? I need something like this: