1. Learn
    2. /
  2. Courses
    3. /
  3. Machine Translation with Keras
  • 1

    Introduction to Machine Translation

    Free

    In this chapter, you'll understand what the encoder-decoder architecture is and how it is used for machine translation. You will also learn about Gated Recurrent Units (GRUs) and how they are used in the encoder-decoder architecture.

  • 2

    Implementing an Encoder-Decoder Model with Keras

    In this chapter, you will implement the encoder-decoder model with the Keras functional API. While doing so, you will learn several useful Keras layers such as RepeatVector and TimeDistributed layers.

  • 3

    Training and Generating Translations

    In this chapter, you will train the previously defined model and then use a well-trained model to generate translations. You will see that our model does a good job when translating sentences.

  • 4

    Teacher Forcing and Word Embeddings

    In this chapter, you will learn about a technique known as Teacher Forcing, which enables translation models to be trained better and faster. Then you will learn how you can use word embeddings to make the model even better.

Initializing

Exercise

Understanding sequential model output

In this exercise you will learn to use the keras.layers.GRU layer. keras.layers.GRU nicely wraps the functionality of a GRU to a Layer object.

You will explore what the shape of the output of a GRU layer looks like and how it changes when different arguments are provided. It is rare to view the numerical vectors produced by a GRU in real life, but in order to use these layers in more complex models, you need to have a good understanding of the shapes of the outputs and how to get the desired output using various arguments.

Here you will have keras, and numpy (as np) loaded already. You can access layers by calling keras.layers.<Layer> or a model by calling keras.models.Model.

Instructions 1/3

undefined XP
    1
    2
    3
  • Create an Input layer of batch size 3, 20 time steps and 5 dimensions and call it inp.
  • Create a GRU layer of hidden size 10, pass the inp to this layer and print the shape of the output.