TensorFlow 2 and Keras in detail

We introduced the general architecture of TensorFlow and trained our first model using Keras. Let's now walk through the main concepts of TensorFlow 2. We will detail several core concepts of TensorFlow, necessary throughout this book, followed by some advanced notions. While we may not employ all of them in the remainder of the book, the readers might find it useful to understand some open source models available on GitHub or to get a deeper understanding of the library.

Core concepts

Released in spring 2019, the new version of the framework focused on simplicity and ease of use. In this section, we will introduce the concepts that TensorFlow relies on and cover how they evolved from version 1 to version 2.

Introducing tensors

TensorFlow takes its name from a mathematical object called a tensor. You can picture tensors as N-dimensional arrays. A tensor could be a scalar, a vector, a 3D matrix, or an N-dimensional matrix.

A fundamental component of TensorFlow...