[PYTHON] When using tf.print (), the contents of the tensor cannot be displayed if it is inside f-string.

Overview

As the title says. It's best not to use f-string intf.print ().

In fact, let's run the following code and compare the output.

import numpy as np
import tensorflow as tf

@tf.function
def add_a_b(a, b):
    c = tf.add(a, b)
    print("(1):", c)
    tf.print("(2):", f"{c}")
    tf.print("(3):", c)
    return c

a = tf.constant(np.asarray([[1, 2], [3, 4]]), dtype=tf.float32)
b = tf.constant(np.asarray([[4, 3], [2, 1]]), dtype=tf.float32)

c = add_a_b(a, b)

The output looks like this. You can see that (2) is f-string but does not benefit fromtf.print ().

(1): Tensor("Add:0", shape=(2, 2), dtype=float32)
(2): Tensor("Add:0", shape=(2, 2), dtype=float32)
(3): [[5 5]
 [5 5]]