Lambda

Lambda(
   *layers, fn, n_units = None, var_list = None, dtype = None, shape = None,
   name = 'lambda', apply_to_layer = False, **kwargs
)

Custom Function Layer

Attributes

tensor : the tensor to be wrapped by this layer
var_list : if vars are involved in the output tensor, they can be specified here
n_units : number of units for this layer,
batch_size : Optional batch size for this layer
apply_to_layer (bool) : if False applies the function to the tensors otherwise applies to the layer and will be listed in variables

Creates a layer from a given tensor that one can then integrate with other layers

Args

layers (Sequence[Layer]) : sequence of input layers
n_units (int) : number of output units (outer dimension)
var_list (List[tf.Variable]) : list of variables used by this Layer
dtype (tf.Dtype) : tensor data type
name (str) : layer name
apply_to_layer (bool) : if True applies function to a layer object else applies the function to the output of previous layer.compute

Methods:

.init_state()

.compute_shape()

.compute(
   *input_tensors
)

.reuse_with(
   *layers, name = None
)