Home > Back-end >  Reducing the dimensions of a tensor in tensorflow
Reducing the dimensions of a tensor in tensorflow

Time:10-22

I have a tensor with the following shape:

> tf.Tensor: shape=(1, 1440)

How do I reduce such a shape so that I can get the following:

> tf.Tensor: shape=(1440,)

CodePudding user response:

Use tf.squeeze:

import tensorflow as tf

tensor = tf.random.uniform((1, 1440))
print(tensor.shape
TensorShape([1, 1440])

And now:

squeezed = tf.squeeze(tensor)
print(squeezed.shape)
TensorShape([1440])

If you really want the comma format for the shape, turn it to NumPy:

tf.squeeze(s).numpy().shape
(1440,)
  • Related