Introduction to language models

Recurrent Neural Networks (RNNs) for Language Modeling with Keras

David Cecchini

Data Scientist

Sentence probability

Many available models

  • Probability of "I loved this movie".
  • Unigram
    • $$P(\text{sentence}) = P(\text{I})P(\text{loved})P(\text{this})P(\text{movie})$$
  • N-gram
    • N = 2 (bigram): $$P(\text{sentence}) = P(\text{I})P(\text{loved} | \text{I})P(\text{this} | \text{loved})P(\text{movie} | \text{this})$$
    • N = 3 (trigram): $$P(\text{sentence}) = P(\text{I})P(\text{loved} | \text{I})P(\text{this} | \text{I loved})P(\text{movie} | \text{loved this})$$
Recurrent Neural Networks (RNNs) for Language Modeling with Keras

Sentence probability (cont.)

  • Skip gram
    • $$P(\text{sentence}) = P(\text{context of I} | \text{I})P(\text{context of loved} | \text{loved}) \ $$ $$P(\text{context of this} | \text{this})P(\text{context of movie} | \text{movie})$$
  • Neural Networks
    • The probability of the sentence is given by a softmax function on the output layer of the network
Recurrent Neural Networks (RNNs) for Language Modeling with Keras

Link to RNNs

Language models are everywhere in RNNs!

  • The network itself

RNN models can be seen as a language model, since it can be used to predict the next word.

Recurrent Neural Networks (RNNs) for Language Modeling with Keras

Link to RNN (cont.)

  • Embedding layer

Displays a macro representation of the layers in a model. The embedding layer need to be the first layer after the input layer, and generates a dense representation of the words.

Recurrent Neural Networks (RNNs) for Language Modeling with Keras

Building vocabulary dictionaries

# Get unique words
unique_words = list(set(text.split(' ')))

# Create dictionary: word is key, index is value
word_to_index = {k:v for (v,k) in enumerate(unique_words)}

# Create dictionary: index is key, word is value
index_to_word = {k:v for (k,v) in enumerate(unique_words)}
Recurrent Neural Networks (RNNs) for Language Modeling with Keras

Preprocessing input

# Initialize variables X and y
X = []
y = []

# Loop over the text: length `sentence_size` per time with step equal to `step`
for i in range(0, len(text) - sentence_size, step):

    X.append(text[i:i + sentence_size])
    y.append(text[i + sentence_size])

# Example (numbers are numerical indexes of vocabulary):
# Sentence is: "i loved this movie" -> (["i", "loved", "this"], "movie")
X[0],y[0] = ([10, 444, 11], 17)
Recurrent Neural Networks (RNNs) for Language Modeling with Keras

Transforming new texts

# Create list to keep the sentences of indexes
new_text_split = []

# Loop and get the indexes from dictionary
for sentence in new_text:

    sent_split = []

    for wd in sentence.split(' '):

        ix = wd_to_index[wd]

        sent_split.append(ix)

    new_text_split.append(sent_split)
Recurrent Neural Networks (RNNs) for Language Modeling with Keras

Let's practice!

Recurrent Neural Networks (RNNs) for Language Modeling with Keras

Preparing Video For Download...