Generation of sequences out of kernel titles

generate_sequences.py

# Generate sequences out of titles:

# Define sequence length
sequence_length = 3

# Generate sequences
def generate_sequences(titles):
    sequences = []
    targets = []
    # Loop for all selected titles
    for title in titles:
        # Run through each title
        if clean_title(title) != '' and clean_title(title) != ' ':
            words = extract_words(title)
            words.append(end_of_sentence)

            for i in range(0, len(words) - sequence_length):
                sequence = words[i:i + sequence_length]
                target = words[i + sequence_length:i + sequence_length + 1]

                sequence_tensor = sequenceToTensor(sequence)
                target_tensor = sequenceToTensor(target)

                sequences.append(sequence_tensor)
                targets.append( target_tensor)
            
    return sequences, targets
    
# generate sequences. I took only 5000 top popular kernel titles
sequences, targets = generate_sequences(popular_kernel_titles[:5000])