3. TensorFlow Development

Activity 3.01: Using TensorBoard to Visualize Tensor Transformations

Solution:

1. Import the TensorFlow library and set a seed:

   import tensorflow as tf
   tf.random.set_seed(42)

2. Set the log directory and initialize a file writer object to write the trace:

   logdir = 'logs/'
   writer = tf.summary.create_file_writer(logdir)

3. Create a TensorFlow function to multiply two tensors and add a value of 1 to all elements in the resulting tensor using the ones_like function to create a tensor of the same shape as the result of the matrix multiplication. Then, apply a sigmoid function to each value of the tensor:

   @tf.function
   def my_func(x, y):
       r1 = tf.matmul(x, y)
       r2 = r1 + tf.ones_like(r1)
       r3 = tf.keras.activations.sigmoid(r2)
       return r3

4. Create two tensors with the shape 5x5x5:

   x = tf.random.uniform((5, 5, 5))
   y = tf.random.uniform((5, 5, 5))

5. Turn...