Updating TensorFlow variables in Python

Python TensorFlow Basic: Exercise-12 with Solution

Write a Python program that initializes a TensorFlow variable with an initial value and then updates its value within a loop. Print the variable's value after each update.

Sample Solution:

Python Code:

import tensorflow as tf

# Create a TensorFlow variable with an initial value
x = tf.Variable(100.0, dtype=tf.float32)

# Define the number of updates
num_updates = 10
# Perform updates within a loop
for i in range(num_updates):
    # Update the variable (e.g., double its value)
    updated_value = x * 2.0
    x.assign(updated_value)
    
    # Print the updated value
    print(f"Update {i + 1}: x = {x.numpy()}")

# Print the final value of the variable
print("Final Variable Value:", x.numpy())

Output:

Update 1: x = 200.0
Update 2: x = 400.0
Update 3: x = 800.0
Update 4: x = 1600.0
Update 5: x = 3200.0
Update 6: x = 6400.0
Update 7: x = 12800.0
Update 8: x = 25600.0
Update 9: x = 51200.0
Update 10: x = 102400.0
Final Variable Value: 102400.0

Explanation:

In the exercise above -

• Initialization: We start by creating a TensorFlow variable named 'x' with an initial value of 10.0. This variable can hold and update floating-point values.
• Number of Updates: Define 'num_updates' as 5, which represents the number of times e want to update the variable.
• Updating in a Loop: Use a for loop to update the variable within each iteration. In this case, we double the current value of the variable (x * 2.0) and assign the updated value back to the variable using variable_tensor.assign(updated_value).
• Printing Updated Values: After each update within the loop, we print the updated value of the variable to observe how it changes.
• Final Value: Finally, after all updates are completed, we print the final value of the variable to see the result of multiple updates.

Python Code Editor:

Previous: Updating TensorFlow variables within a session in Python.
Next: Common TensorFlow data types in Python.