2.5 Graphs and Sessions: Creating and Running a TensorFlow Session

Creating and running a TensorFlow session is a fundamental aspect of TensorFlow usage, encapsulating the environment for executing operations within the computational graph. In TensorFlow 1.x, sessions are created using tf.Session() class, whereas in TensorFlow 2.x with eager execution, sessions are not explicitly required as operations are executed immediately. Executing operations within a session involves using sess.run() method, passing tensors as arguments to compute their output. In TensorFlow 1.x, sessions are typically managed within a context manager to ensure automatic closure and resource release. However, in TensorFlow 2.x, eager execution simplifies this process by managing resources automatically. Mastering session creation and management is essential for effectively utilizing TensorFlow, enabling efficient execution of operations and computation of tensor outputs within the computational graph.