autoencoder_model: Autoencoder keras model wrapper over simple dense layers
In ifrit98/layerR:

Autoencoder keras model wrapper over simple dense layers

autoencoder_model(
  original_dim,
  hidden_dims,
  num_layers = 3,
  regularizer = NULL,
  use_batchnorm = FALSE,
  name = NULL
)

`original_dim`	integer denoting original dimensionality of input space
`hidden_dims`	may be an integer, or a list containing desired units for encoder step. 'list(512, 256, 64)' will give a 'code_dimension' of '64'
`num_layers`	integer, number of layers to use in each encoder decoder. [num_layers] must equal [length(hidden_dims)] unless hidden_dims is a scalar, in which case, hidden_dims is automatically computed based on nearest powers of 2
`regularizer`	callable or string name of keras regularizer
`use_batchnorm`	bool indicating use of batchnorm layer inbetween each dense
`name`	(optional) string to name the model