tf.squeeze  |  TensorFlow v2.16.1 (original) (raw)

tf.squeeze

Stay organized with collections Save and categorize content based on your preferences.

Removes dimensions of size 1 from the shape of a tensor.

tf.squeeze(
    input, axis=None, name=None
)

Used in the notebooks

Given a tensor input, this operation returns a tensor of the same type with all dimensions of size 1 removed. If you don't want to remove all size 1 dimensions, you can remove specific size 1 dimensions by specifyingaxis.

For example:

# 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
tf.shape(tf.squeeze(t))  # [2, 3]

Or, to remove specific size 1 dimensions:

# 't' is a tensor of shape [1, 2, 1, 3, 1, 1]
tf.shape(tf.squeeze(t, [2, 4]))  # [1, 2, 3, 1]

Unlike the older op tf.compat.v1.squeeze, this op does not accept a deprecated squeeze_dims argument.

@tf.function
def func(x):
  print('x.shape:', x.shape)
  known_axes = [i for i, size in enumerate(x.shape) if size == 1]
  y = tf.squeeze(x, axis=known_axes)
  print('shape of tf.squeeze(x, axis=known_axes):', y.shape)
  y = tf.squeeze(x)
  print('shape of tf.squeeze(x):', y.shape)
  return 0

_ = func.get_concrete_function(tf.TensorSpec([None, 1, 2], dtype=tf.int32))
# Output is.
# x.shape: (None, 1, 2)
# shape of tf.squeeze(x, axis=known_axes): (None, 2)
# shape of tf.squeeze(x): <unknown>