TensorFlow: AttributeError: 'Tensor' object has no attribute 'shape'

Question

I have the following code which uses TensorFlow. After I reshape a list, it says

AttributeError: 'Tensor' object has no attribute 'shape'

when I try to print its shape.

# Get the shape of the training data.
print "train_data.shape: " + str(train_data.shape)
train_data = tf.reshape(train_data, [400, 1])
print "train_data.shape: " + str(train_data.shape)
train_size,num_features = train_data.shape

Output:

train_data.shape: (400,) Traceback (most recent call last): File "", line 1, in File "/home/shehab/Downloads/tools/python/pycharm-edu-2.0.4/helpers/pydev/pydev_import_hook.py", line 21, in do_import module = self._system_import(name, *args, **kwargs) File "/home/shehab/Dropbox/py-projects/try-tf/logistic_regression.py", line 77, in print "train_data.shape: " + str(train_data.shape) AttributeError: 'Tensor' object has no attribute 'shape'

Could anyone please tell me what I am missing?

Answer 1

UPDATE: Since TensorFlow 1.0, tf.Tensor now has a tf.Tensor.shape property, which returns the same value as tf.Tensor.get_shape().

---

Indeed, in versions prior to TensorFlow 1.0 tf.Tensor doesn't have a .shape property. You should use the Tensor.get_shape() method instead:

train_data = tf.reshape(train_data, [400, 1])
print "train_data.shape: " + str(train_data.get_shape())

Note that in general you might not be able to get the actual shape of the result of a TensorFlow operation. In some cases, the shape will be a computed value that depends on running the computation to find its value; and it may even vary from one run to the next (e.g. the shape of tf.unique()). In that case, the result of get_shape() for some dimensions may be None (or "?").

Answer 2

import tensorflow as tf

and replace train_data.shape with tf.Session.run(tf.rank(train_data))