tf.types.experimental.AtomicFunction

Base class for graph functions.

Inherits From: Callable

An AtomicFunction encapsulates a single graph function definition.

AtomicFunction can be called directly only if no captures are needed according to the FunctionType. If captures are present, please use call_with_captures instead.

AtomicFunction does not support gradients. Please use the parent ConcreteFunction if you need gradient support.

Attributes
function_type	Returns a FunctionType describing this callable.

Methods

call_with_captures

View source

call_with_captures(
    args, kwargs, captures
)

Calls this AtomicFunction with captures as defined by its FunctionType.

Args
args	Tuple containing positional arguments
kwargs	Dict containing keyword arguments
captures	Tuple of tensors supplying captured tensor values.

Returns
A structured output value based on the inputs.

__call__

View source

__call__(
    *args, **kwargs
)

Executes this callable.

This behaves like a regular op - in eager mode, it immediately starts execution, returning results. In graph mode, it creates ops which return symbolic TensorFlow values (like tf.Tensor, tf.data.Dataset, etc.). For example, tf.function callables typically generate a tf.raw_ops.PartitionedCall op, but not always - the exact operations being generated are an internal implementation detail.

Args
*args	positional argument for this call
**kwargs	keyword arguments for this call

Returns
The execution results.