73、TensorFlow Graphs: An In - Depth Exploration

TensorFlow Graphs: An In - Depth Exploration

1. TF Functions and Concrete Functions

TF Functions are polymorphic, supporting inputs of different types and shapes. Consider the following tf_cube() function:

import tensorflow as tf

@tf.function 
def tf_cube(x): 
    return x ** 3

Every time a TF Function is called with a new combination of input types or shapes (input signature), a new concrete function is generated. For example:

concrete_function = tf_cube.get_concrete_function(tf.constant(2.0)) 
print(concrete_function) 
print(concrete_function(tf.constant(2.0))) 

The tensors in the graphs generated by these functions are symbol