Chapter 3 Getting Started with TensorFlow
The three elements of TensorFlow: computational model, data model, and running model
TensorFlow Computational Model - Computational Graph
Tensor: Tensor, which can be simply understood as a multidimensional array
Flow: flow
TensorFlow programs can generally be divided into two phases. In the first stage all computations in the computation graph need to be defined. The second stage is the execution stage.
Example of definition stage:
During this process, TensorFlow will automatically convert the defined computations into nodes on the computation graph. In a TensorFlow program, the system automatically maintains a default computation graph.
The current default calculation graph can be obtained through the tf.get_default_graph function.