Tensor to NumPy: NumPy Array To TensorFlow Tensor And Back

Video Transcript

Today, we’re going to learn how to convert between NumPy arrays and TensorFlow tensors and back.

We’re going to begin by creating a file: numpy-arrays-to-tensorflow-tensors-and-back.py.

# command line
e numpy-arrays-to-tensorflow-tensors-and-back.py

We’re going to import TensorFlow as tf, that’s the standard, and import NumPy as np.

# numpy-arrays-to-tensorflow-tensors-and-back.py file
import tensorflow as tf
import numpy as np

We’re going to begin by generating a NumPy array by using the random.rand method to generate a 3 by 2 random matrix using NumPy.

# numpy-arrays-to-tensorflow-tensors-and-back.py file
# ...
np_array = np.random.rand(3, 2)

If we run the code, we can see that it’s just a standard NumPy array.

# command line
python numpy-arrays-to-tensorflow-tensors-and-back.py

But then I’m going to create a standard TensorFlow session using the tf.Session method.

# numpy-arrays-to-tensorflow-tensors-and-back.py file
# ...
sess = tf.Session()

We’re then going to use a session to convert the NumPy array into TensorFlow tensor using the tf.constant method.

# numpy-arrays-to-tensorflow-tensors-and-back.py file
# ...
with sess.as_default():
    tensor = tf.constant(np_array)

And then we’re going to print that tensor

# numpy-arrays-to-tensorflow-tensors-and-back.py file
# ...
    print(tensor)

so you’re going to see that it actually is a TensorFlow tensor.

# command line
python numpy-arrays-to-tensorflow-tensors-and-back.py

As you see, it is a tensor.

Finally, we’re going to convert it back by using the tensor.eval method.

# numpy-arrays-to-tensorflow-tensors-and-back.py file
# ...
    numpy_array_2 = tensor.eval()
    print(numpy_array_2)

As you can see, we’ve got our original NumPy array back.

# command line
python numpy-arrays-to-tensorflow-tensors-and-back.py