Keras is another popularly used deep learning library. It was developed by François Chollet at Google. It is well known for its fast prototyping, and it makes model building simple. It is a high-level library, meaning that it does not perform any low-level operations on its own, such as convolution. It uses a backend engine for doing that, such as TensorFlow. The Keras API is available in tf.keras, and TensorFlow 2.0 uses it as the primary API.

Building a model in Keras involves four important steps:

1. Defining the model
2. Compiling the model
3. Fitting the model
4. Evaluating the model