Search code examples
pythontensorflowtensorflow2.0tensorflow2.x

Tensorflow 2.0.0: AttributeError: 'TensorSliceDataset' object has no attribute 'as_numpy_iterator'

I am testing tensorflow tf.data.Dataset method as_numpy_iterator using tensorflow 2.0.0. According to the official documentation https://www.tensorflow.org/api_docs/python/tf/data/Dataset?version=stable#as_numpy_iterator, this function allows directly inspecting the content of a tensorflow dataset. But when I try the given example:

dataset = tf.data.Dataset.from_tensor_slices([1, 2, 3]) 
for element in dataset.as_numpy_iterator(): 
  print(element)

There occurs an error: AttributeError: 'TensorSliceDataset' object has no attribute 'as_numpy_iteractor'. I am wondering if this method is just newly added, beyond the support of tensorflow 2.0.0. If so, is there an alternative to checking the dataset content as the as_numpy_iterator()?

Solution

  • The link to the documentation that you provided points to

    TensorFlow Core r2.1

    Updating your tensorflow version to version 2.1 should solve the issue;

    The method .as_numpy_iterator() is not present in TensorFlow 2.0, but only in TensorFlow >= 2.1