tf.data.IteratorSpec

Type specification for tf.data.Iterator.

Inherits From: TypeSpec

For instance, tf.data.IteratorSpec can be used to define a tf.function that takes tf.data.Iterator as an input argument:

@tf.function(input_signature=[tf.data.IteratorSpec(
  tf.TensorSpec(shape=(), dtype=tf.int32, name=None))])
def square(iterator):
  x = iterator.get_next()
  return x * x
dataset = tf.data.Dataset.from_tensors(5)
iterator = iter(dataset)
print(square(iterator))
tf.Tensor(25, shape=(), dtype=int32)

element_spec A (nested) structure of tf.TypeSpec objects that represents the type specification of the iterator elements.
value_type The Python type for values that are compatible with this TypeSpec.

In particular, all values t