Have a question? Connect with the community at the TensorFlow Forum Visit Forum


Returns the constant value of the given tensor, if efficiently calculable.

Used in the notebooks

Used in the tutorials

This function attempts to partially evaluate the given tensor, and returns its value as a numpy ndarray if this succeeds.

Example usage:

a = tf.constant(10)
b = tf.constant(20)
tf.get_static_value(tf.add(a, b))
# `tf.Variable` is not supported.
c = tf.Variable(30)

Using partial option is most relevant when calling get_static_value inside a tf.function. Setting it to True will return the results but for the values that cannot be evaluated will be None. For example:

class Foo(object):
  def __init__(self):
    self.a = tf.Variable(1)
    self.b = tf.constant(2)

  def bar(self, partial):
    packed = tf.raw_ops.Pack(values=[self.a, self.b])
    static_val = tf.get_static_value(packed, partial=partial)

f = Foo()
f.bar(partial=True)  # `array([None, array(2, dtype=int32)], dtype=object)`
f.bar(partial=False)  # `None`

Compatibility(V1): If constant_value(tensor) returns a non-None result, it will no longer be possible to feed a different value for tensor. This allows the result of this function to influence the graph that is constructed, and permits static shape optimizations.

tensor The Tensor to be evaluated.
partial If True, the returned numpy array is allowed to have partially evaluated values. Values that can't be evaluated will be None.

A numpy ndarray containing the constant value of the given tensor, or None if it cannot be calculated.

TypeError if tensor is not an ops.Tensor.