Graphs
One of the most useful features of TensorFlow is the ability to visualize your models using graphs. In this tutorial, we will cover the basics of understanding graphs in TensorFlow.
What is a Graph in TensorFlow?
-
A TensorFlow graph is a representation of a computation in TensorFlow.
-
The graph is a combination of nodes and edges. nodes represent mathematical operations, and edges represent the flow of data between the nodes.
-
Nodes are typically operations such as matrix multiplication, addition etc.
-
Edges represent the tensors, or data, that flow between the operations.
Advantages of Graph
-
TensorFlow graphs allow for complex computations to be broken down into smaller, more manageable parts.
-
Graphs in TensorFlow are optimized for execution on a variety of hardware, including CPUs andGPUs.
-
TensorFlow graphs can be visualized using tools like TensorBoard.
Example
import tensorflow as tf
graph = tf.Graph()
with graph.as_default():
a = tf.constant(2.0, name="input_a")
b = tf.constant(3.0, name="input_b")
c = tf.add(a, b, name="sum_c")
with tf.compat.v1.Session(graph=graph) as sess:
result = sess.run(c)
print(result)
Computation
C = a + b
Graph
