Preparing the data for training

In this exercise, you will continue to prepare the data to train the model. After creating the arrays of sentences and next characters, you need to transform them to numerical values that can be used on the model.

This step is necessary because the RNN models expect numbers only and not strings. You will create numerical arrays that have zeros or ones in the positions representing the characters present on the sentences. Ones (or True) represent the corresponding character is present, while zeros (or False) represent the absence of the character in that position of the sentence.

The variables sentences, next_char, n_vocab, chars_window, num_seqs (number of sentences in the training data) are already loaded in the environment, as well as numpy as np.

Instantiate a np.array() with zeros and shape (number of sentences, characters window, vocabulary size).
Use the dictionary char_to_index to set the position of the current char to 1.
Set the current next character to 1.
Print the first position of each array.

Exercise

Preparing the data for training

Instructions