1. Apprendre
    2. /
  2. Cours
    3. /
  3. Recurrent Neural Networks (RNNs) for Language Modeling with Keras
Connected

Exercice

Creating the text generation model

In this exercise, you will define a text generation model using Keras.

The variables n_vocab containing the vocabulary size and input_shape containing the shape of the data used for training are already loaded in the environment. Also, the weights of a pre-trained model is available on file model_weights.h5. The model was trained with 40 epochs on the training data. Recap that to train a model in Keras, you just use the method .fit() on the training data (X, y), and the parameter epochs. For example:

model.fit(X_train, y_train, epochs=40)

Instructions

100 XP
  • Add one LSTM layer returning the sequences.
  • Add one LSTM layer not returning the sequences.
  • Add the output layer with n_vocab units.
  • Display the model summary.