Class FuncGraph

tensorflow/python/framework/func_graph.py:140–733

Graph representing a function body. Attributes: name: The name of the function. inputs: Placeholder tensors representing the inputs to this function. The tensors are in this FuncGraph. This represents "regular" inputs as well as captured inputs (i.e. the values of self.capture

class FuncGraph(ops.Graph):
  """Graph representing a function body.

  Attributes:
    name: The name of the function.
    inputs: Placeholder tensors representing the inputs to this function. The
      tensors are in this FuncGraph. This represents "regular" inputs as well as
      captured inputs (i.e. the values of self.captures), with the regular
      inputs coming first.
    outputs: Tensors that will be returned by this function. The tensors are in
      this FuncGraph.
    control_outputs: Operations that must be executed before the function
      represented by this graph can be said to have been executed.
    structured_input_signature: A tuple of (args, kwargs), which are both
      possibly-nested python objects that were received by this function. Note
      that these structures might contain Python `None`s.
    structured_outputs: A possibly-nested python object which will be returned
      by this function. The Tensors in this structure are the same as those of
      self.outputs. Note that this structure might contain Python `None`s.
    variables: Variables that should be watched during function execution.
    outer_graph: The graph this function is defined in. May be another FuncGraph
      or the global default Graph.
    captures: Maps external tensor -> internal tensor (i.e. input placeholder).
      The entries are in the order they were captured.
    control_captures: Set of external ops on which this graph has a control
      dependency.
    seed: The graph-level random seed.
    capture_by_value: If True, the func graph will capture Variables by value
      instead of reference.
  """

  def __init__(self, name, collections=None, capture_by_value=None):
    """Construct a new FuncGraph.

    The graph will inherit its graph key, collections, seed, and distribution
    strategy stack from the current context or graph.

    Args:
      name: the name of the function.
      collections: a dictionary of collections this FuncGraph should start
        with. If not specified (None), the FuncGraph will read (but not write
        to) the outer graph's collections that are not whitelisted, and both
        read and write to the outer graph's collections that are whitelisted.
        The current whitelisted collections are the global variables, the
        local variables, and the trainable variables.
        Defaults to None.
      capture_by_value: An optional boolean. If True, the func graph will
        capture Variables by value instead of reference. By default inherit
        from outer graphs, and failing that will default to False.
    """
    super(FuncGraph, self).__init__()

    self.name = name
    self.inputs = []
    self.outputs = []
    self.control_outputs = []
    self.control_captures = set()
    self.structured_input_signature = None