Wednesday, May 31, 2017

Bot for Generating TV Scripts-- RNNs

dlnd_tv_script_generation

TV Script Generation

In this project, we'll generate our own Simpsons TV scripts using RNNs. we'll be using part of the Simpsons dataset of scripts from 27 seasons. The Neural Network we'll build will generate a new TV script for a scene at Moe's Tavern.

Get the Data

The data is already provided . we'll be using a subset of the original dataset. It consists of only the scenes in Moe's Tavern. This doesn't include other versions of the tavern, like "Moe's Cavern", "Flaming Moe's", "Uncle Moe's Family Feed-Bag", etc..

In [1]:
import helper

data_dir = './data/simpsons/moes_tavern_lines.txt'
text = helper.load_data(data_dir)
# Ignore notice, since we don't use it for analysing the data
text = text[81:]
#text

Explore the Data

Play around with view_sentence_range to view different parts of the data.

In [2]:
view_sentence_range = (0, 10)


import numpy as np

print('Dataset Stats')
print('Roughly the number of unique words: {}'.format(len({word: None for word in text.split()})))
scenes = text.split('\n\n')
print('Number of scenes: {}'.format(len(scenes)))
sentence_count_scene = [scene.count('\n') for scene in scenes]
print('Average number of sentences in each scene: {}'.format(np.average(sentence_count_scene)))

sentences = [sentence for scene in scenes for sentence in scene.split('\n')]
print('Number of lines: {}'.format(len(sentences)))
word_count_sentence = [len(sentence.split()) for sentence in sentences]
print('Average number of words in each line: {}'.format(np.average(word_count_sentence)))

print()
print('The sentences {} to {}:'.format(*view_sentence_range))
print('\n'.join(text.split('\n')[view_sentence_range[0]:view_sentence_range[1]]))
Dataset Stats
Roughly the number of unique words: 11492
Number of scenes: 262
Average number of sentences in each scene: 15.248091603053435
Number of lines: 4257
Average number of words in each line: 11.50434578341555

The sentences 0 to 10:
Moe_Szyslak: (INTO PHONE) Moe's Tavern. Where the elite meet to drink.
Bart_Simpson: Eh, yeah, hello, is Mike there? Last name, Rotch.
Moe_Szyslak: (INTO PHONE) Hold on, I'll check. (TO BARFLIES) Mike Rotch. Mike Rotch. Hey, has anybody seen Mike Rotch, lately?
Moe_Szyslak: (INTO PHONE) Listen you little puke. One of these days I'm gonna catch you, and I'm gonna carve my name on your back with an ice pick.
Moe_Szyslak: What's the matter Homer? You're not your normal effervescent self.
Homer_Simpson: I got my problems, Moe. Give me another one.
Moe_Szyslak: Homer, hey, you should not drink to forget your problems.
Barney_Gumble: Yeah, you should only drink to enhance your social skills.


Implement Preprocessing Functions

The first thing to do to any dataset is preprocessing. Implement the following preprocessing functions below:

  • Lookup Table
  • Tokenize Punctuation

Lookup Table

To create a word embedding, we first need to transform the words to ids. In this function, we'll create two dictionaries:

  • Dictionary to go from the words to an id, we'll call vocab_to_int
  • Dictionary to go from the id to word, we'll call int_to_vocab

we'll Return these dictionaries in the following tuple (vocab_to_int, int_to_vocab)

In [3]:
import numpy as np
import problem_unittests as tests

def create_lookup_tables(text):
    """
    Create lookup tables for vocabulary
    :param text: The text of tv scripts split into words
    :return: A tuple of dicts (vocab_to_int, int_to_vocab)
    """
    # TODO: Implement Function
    vocab=set(text)
    vocab_to_int={word:i for i,word in enumerate(vocab)}
    int_to_vocab=dict(enumerate(vocab))
    
    return vocab_to_int, int_to_vocab



tests.test_create_lookup_tables(create_lookup_tables)
Tests Passed

Tokenize Punctuation

We'll be splitting the script into a word array using spaces as delimiters. However, punctuations like periods and exclamation marks make it hard for the neural network to distinguish between the word "bye" and "bye!".

Implement the function token_lookup to return a dict that will be used to tokenize symbols like "!" into "||Exclamation_Mark||". Create a dictionary for the following symbols where the symbol is the key and value is the token:

  • Period ( . )
  • Comma ( , )
  • Quotation Mark ( " )
  • Semicolon ( ; )
  • Exclamation mark ( ! )
  • Question mark ( ? )
  • Left Parentheses ( ( )
  • Right Parentheses ( ) )
  • Dash ( -- )
  • Return ( \n )

This dictionary will be used to token the symbols and add the delimiter (space) around it. This separates the symbols as it's own word, making it easier for the neural network to predict on the next word. Make sure you don't use a token that could be confused as a word. Instead of using the token "dash", try using something like "||dash||".

In [4]:
def token_lookup():
    """
    Generate a dict to turn punctuation into a token.
    :return: Tokenize dictionary where the key is the punctuation and the value is the token
    """
    # TODO: Implement Function
    tokenize_punct={".":"||Period||",
                   ",":"||Comma||",
                   "\"":"||Quotation_Mark||",
                   ";":"||SemiColon||",
                   "!":"||Exclamation_Mark||",
                   "?":"||Question_Mark||",
                   "(":"||Left_Parenthesis||",
                   ")":"||Right_Parenthesis||",
                   "--":"||Dash||",
                   "\n":"||Return||"}
    return tokenize_punct


tests.test_tokenize(token_lookup)
Tests Passed

Preprocess all the data and save it

Running the code cell below will preprocess all the data and save it to file.

In [5]:
# Preprocess Training, Validation, and Testing Data
helper.preprocess_and_save_data(data_dir, token_lookup, create_lookup_tables)

Check Point

This is your first checkpoint. If you ever decide to come back to this notebook or have to restart the notebook, you can start from here. The preprocessed data has been saved to disk.

In [6]:
import helper
import numpy as np
import problem_unittests as tests

int_text, vocab_to_int, int_to_vocab, token_dict = helper.load_preprocess()

Build the Neural Network

You'll build the components necessary to build a RNN by implementing the following functions below:

  • get_inputs
  • get_init_cell
  • get_embed
  • build_rnn
  • build_nn
  • get_batches

Check the Version of TensorFlow and Access to GPU

In [7]:
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer'
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))
TensorFlow Version: 1.0.0
Default GPU Device: /gpu:0

Input

Implement the get_inputs() function to create TF Placeholders for the Neural Network. It should create the following placeholders:

  • Input text placeholder named "input" using the TF Placeholder name parameter.
  • Targets placeholder
  • Learning Rate placeholder

Return the placeholders in the following the tuple (Input, Targets, LearingRate)

In [8]:
def get_inputs():
    """
    Create TF Placeholders for input, targets, and learning rate.
    :return: Tuple (input, targets, learning rate)
    """
    # TODO: Implement Function
    inputs=tf.placeholder(tf.int32,[None,None],name="input")
    targets=tf.placeholder(tf.int32,[None,None],name="targets")
    learning_rate=tf.placeholder(tf.float32,None,name="learning_rate")
    return inputs, targets, learning_rate



tests.test_get_inputs(get_inputs)
Tests Passed

Build RNN Cell and Initialize

Stack one or more BasicLSTMCells in a MultiRNNCell.

  • The Rnn size should be set using rnn_size
  • Initalize Cell State using the MultiRNNCell's zero_state() function
    • Apply the name "initial_state" to the initial state using tf.identity()

Return the cell and initial state in the following tuple (Cell, InitialState)

In [9]:
def get_init_cell(batch_size, rnn_size):
    """
    Create an RNN Cell and initialize it.
    :param batch_size: Size of batches
    :param rnn_size: Size of RNNs
    :return: Tuple (cell, initialize state)
    """
    # TODO: Implement Function
    num_layers=2
    lstm_layer=tf.contrib.rnn.BasicLSTMCell(rnn_size)
    cell=tf.contrib.rnn.MultiRNNCell([lstm_layer]*num_layers)
    initial_state = cell.zero_state(batch_size, tf.float32)
    initial_state=tf.identity(initial_state,name="initial_state")
    return cell, initial_state



tests.test_get_init_cell(get_init_cell)
Tests Passed

Word Embedding

Apply embedding to input_data using TensorFlow. Return the embedded sequence.

In [10]:
def get_embed(input_data, vocab_size, embed_dim):
    """
    Create embedding for <input_data>.
    :param input_data: TF placeholder for text input.
    :param vocab_size: Number of words in vocabulary.
    :param embed_dim: Number of embedding dimensions
    :return: Embedded input.
    """
    # TODO: Implement Function
    embedding = tf.Variable(tf.random_uniform((vocab_size, embed_dim), -1, 1))
    embed = tf.nn.embedding_lookup(embedding, input_data)
    return embed



tests.test_get_embed(get_embed)
Tests Passed

Build RNN

You created a RNN Cell in the get_init_cell() function. Time to use the cell to create a RNN.

Return the outputs and final_state state in the following tuple (Outputs, FinalState)

In [11]:
def build_rnn(cell, inputs):
    """
    Create a RNN using a RNN Cell
    :param cell: RNN Cell
    :param inputs: Input text data
    :return: Tuple (Outputs, Final State)
    """
    # TODO: Implement Function
    outputs,final_state=tf.nn.dynamic_rnn(cell,inputs,dtype=tf.float32)
    final_state=tf.identity(final_state,name="final_state")
    return outputs, final_state



tests.test_build_rnn(build_rnn)
Tests Passed

Build the Neural Network

Apply the functions you implemented above to:

  • Apply embedding to input_data using your get_embed(input_data, vocab_size, embed_dim) function.
  • Build RNN using cell and your build_rnn(cell, inputs) function.
  • Apply a fully connected layer with a linear activation and vocab_size as the number of outputs.

Return the logits and final state in the following tuple (Logits, FinalState)

In [12]:
def build_nn(cell, rnn_size, input_data, vocab_size, embed_dim):
    """
    Build part of the neural network
    :param cell: RNN cell
    :param rnn_size: Size of rnns
    :param input_data: Input data
    :param vocab_size: Vocabulary size
    :param embed_dim: Number of embedding dimensions
    :return: Tuple (Logits, FinalState)
    """
    # TODO: Implement Function
    embeddings=get_embed(input_data,vocab_size,embed_dim)
    outputs,final_state=build_rnn(cell,embeddings)
    w_init=tf.truncated_normal_initializer(stddev=0.01)
    b_init=tf.zeros_initializer()
    fully_connected= tf.contrib.layers.fully_connected(outputs,
                                                       vocab_size,
                                                       weights_initializer=w_init,
                                                      biases_initializer=b_init,
                                                      activation_fn=None)
    return fully_connected,final_state



tests.test_build_nn(build_nn)
Tests Passed

Batches

Implement get_batches to create batches of input and targets using int_text. The batches should be a Numpy array with the shape (number of batches, 2, batch size, sequence length). Each batch contains two elements:

  • The first element is a single batch of input with the shape [batch size, sequence length]
  • The second element is a single batch of targets with the shape [batch size, sequence length]

If you can't fill the last batch with enough data, drop the last batch.

For exmple, get_batches([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15], 2, 3) would return a Numpy array of the following:

[
  # First Batch
  [
    # Batch of Input
    [[ 1  2  3], [ 7  8  9]],
    # Batch of targets
    [[ 2  3  4], [ 8  9 10]]
  ],

  # Second Batch
  [
    # Batch of Input
    [[ 4  5  6], [10 11 12]],
    # Batch of targets
    [[ 5  6  7], [11 12 13]]
  ]
]
In [13]:
def get_batches(int_text, batch_size, seq_length):
    """
    Return batches of input and target
    :param int_text: Text with the words replaced by their ids
    :param batch_size: The size of batch
    :param seq_length: The length of sequence
    :return: Batches as a Numpy array
    """
    # TODO: Implement Function
    n_batches= int(len(int_text)//(batch_size*seq_length))
    
    x=np.array(int_text[:n_batches*batch_size*seq_length])
    y=np.array(int_text[1:n_batches*batch_size*seq_length+1])
    #print(x.shape)
    #print(y.shape)

    y[-1]=x[0]

    x_batches=np.array(np.split(x.reshape(batch_size,-1),n_batches,1))
    #print(x_batches)
    y_batches=np.array(np.split(y.reshape(batch_size,-1),n_batches,1))
    
    return np.array(list(zip(x_batches,y_batches)))



tests.test_get_batches(get_batches)
Tests Passed

Neural Network Training

Hyperparameters

Tune the following parameters:

  • Set num_epochs to the number of epochs.
  • Set batch_size to the batch size.
  • Set rnn_size to the size of the RNNs.
  • Set embed_dim to the size of the embedding.
  • Set seq_length to the length of sequence.
  • Set learning_rate to the learning rate.
  • Set show_every_n_batches to the number of batches the neural network should print progress.
In [14]:
# Number of Epochs
num_epochs = 200
# Batch Size
batch_size = 512
# RNN Size
rnn_size = 1024
# Embedding Dimension Size
embed_dim = 500
# Sequence Length
seq_length = 15
# Learning Rate
learning_rate = .001
# Show stats for every n number of batches
show_every_n_batches = 100


save_dir = './save'

Build the Graph

Build the graph using the neural network you implemented.

In [15]:
from tensorflow.contrib import seq2seq

train_graph = tf.Graph()
with train_graph.as_default():
    vocab_size = len(int_to_vocab)
    input_text, targets, lr = get_inputs()
    input_data_shape = tf.shape(input_text)
    cell, initial_state = get_init_cell(input_data_shape[0], rnn_size)
    logits, final_state = build_nn(cell, rnn_size, input_text, vocab_size, embed_dim)

    # Probabilities for generating words
    probs = tf.nn.softmax(logits, name='probs')

    # Loss function
    cost = seq2seq.sequence_loss(
        logits,
        targets,
        tf.ones([input_data_shape[0], input_data_shape[1]]))

    # Optimizer
    optimizer = tf.train.AdamOptimizer(lr)

    # Gradient Clipping
    gradients = optimizer.compute_gradients(cost)
    capped_gradients = [(tf.clip_by_value(grad, -1., 1.), var) for grad, var in gradients]
    train_op = optimizer.apply_gradients(capped_gradients)

Train

Train the neural network on the preprocessed data. If you have a hard time getting a good loss, check the forms to see if anyone is having the same problem.

In [16]:
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
batches = get_batches(int_text, batch_size, seq_length)

with tf.Session(graph=train_graph) as sess:
    sess.run(tf.global_variables_initializer())

    for epoch_i in range(num_epochs):
        state = sess.run(initial_state, {input_text: batches[0][0]})

        for batch_i, (x, y) in enumerate(batches):
            feed = {
                input_text: x,
                targets: y,
                initial_state: state,
                lr: learning_rate}
            train_loss, state, _ = sess.run([cost, final_state, train_op], feed)

            # Show every <show_every_n_batches> batches
            if (epoch_i * len(batches) + batch_i) % show_every_n_batches == 0:
                print('Epoch {:>3} Batch {:>4}/{}   train_loss = {:.3f}'.format(
                    epoch_i,
                    batch_i,
                    len(batches),
                    train_loss))

    # Save Model
    saver = tf.train.Saver()
    saver.save(sess, save_dir)
    print('Model Trained and Saved')
Epoch   0 Batch    0/8   train_loss = 8.821
Epoch  12 Batch    4/8   train_loss = 5.590
Epoch  25 Batch    0/8   train_loss = 4.842
Epoch  37 Batch    4/8   train_loss = 4.344
Epoch  50 Batch    0/8   train_loss = 3.856
Epoch  62 Batch    4/8   train_loss = 3.453
Epoch  75 Batch    0/8   train_loss = 3.023
Epoch  87 Batch    4/8   train_loss = 2.550
Epoch 100 Batch    0/8   train_loss = 2.238
Epoch 112 Batch    4/8   train_loss = 1.769
Epoch 125 Batch    0/8   train_loss = 1.245
Epoch 137 Batch    4/8   train_loss = 0.874
Epoch 150 Batch    0/8   train_loss = 0.710
Epoch 162 Batch    4/8   train_loss = 0.430
Epoch 175 Batch    0/8   train_loss = 0.345
Epoch 187 Batch    4/8   train_loss = 0.257
Model Trained and Saved

Save Parameters

Save seq_length and save_dir for generating a new TV script.

In [17]:
# Save parameters for checkpoint
helper.save_params((seq_length, save_dir))

Checkpoint

In [18]:
import tensorflow as tf
import numpy as np
import helper
import problem_unittests as tests

_, vocab_to_int, int_to_vocab, token_dict = helper.load_preprocess()
seq_length, load_dir = helper.load_params()

Implement Generate Functions

Get Tensors

Get tensors from loaded_graph using the function get_tensor_by_name(). Get the tensors using the following names:

  • "input:0"
  • "initial_state:0"
  • "final_state:0"
  • "probs:0"

Return the tensors in the following tuple (InputTensor, InitialStateTensor, FinalStateTensor, ProbsTensor)

In [19]:
def get_tensors(loaded_graph):
    """
    Get input, initial state, final state, and probabilities tensor from <loaded_graph>
    :param loaded_graph: TensorFlow graph loaded from file
    :return: Tuple (InputTensor, InitialStateTensor, FinalStateTensor, ProbsTensor)
    """
    # TODO: Implement Function
    InputTensor=loaded_graph.get_tensor_by_name('input:0')
    InitialStateTensor=loaded_graph.get_tensor_by_name('initial_state:0')
    FinalStateTensor=loaded_graph.get_tensor_by_name('final_state:0')
    ProbsTensor=loaded_graph.get_tensor_by_name('probs:0')
    return (InputTensor, InitialStateTensor, FinalStateTensor, ProbsTensor)



tests.test_get_tensors(get_tensors)
Tests Passed

Choose Word

Implement the pick_word() function to select the next word using probabilities.

In [20]:
def pick_word(probabilities, int_to_vocab):
    """
    Pick the next word in the generated text
    :param probabilities: Probabilites of the next word
    :param int_to_vocab: Dictionary of word ids as the keys and words as the values
    :return: String of the predicted word
    """
    # TODO: Implement Function
    word_id=np.random.choice(len(probabilities),p=probabilities)
    
    return int_to_vocab[word_id]



tests.test_pick_word(pick_word)
Tests Passed

Generate TV Script

This will generate the TV script for you. Set gen_length to the length of TV script you want to generate.

In [21]:
gen_length = 200
# homer_simpson, moe_szyslak, or Barney_Gumble
prime_word = 'moe_szyslak'


loaded_graph = tf.Graph()
with tf.Session(graph=loaded_graph) as sess:
    # Load saved model
    loader = tf.train.import_meta_graph(load_dir + '.meta')
    loader.restore(sess, load_dir)

    # Get Tensors from loaded model
    input_text, initial_state, final_state, probs = get_tensors(loaded_graph)

    # Sentences generation setup
    gen_sentences = [prime_word + ':']
    prev_state = sess.run(initial_state, {input_text: np.array([[1]])})

    # Generate sentences
    for n in range(gen_length):
        # Dynamic Input
        dyn_input = [[vocab_to_int[word] for word in gen_sentences[-seq_length:]]]
        dyn_seq_length = len(dyn_input[0])

        # Get Prediction
        probabilities, prev_state = sess.run(
            [probs, final_state],
            {input_text: dyn_input, initial_state: prev_state})
        
        pred_word = pick_word(probabilities[dyn_seq_length-1], int_to_vocab)

        gen_sentences.append(pred_word)
    
    # Remove tokens
    tv_script = ' '.join(gen_sentences)
    for key, token in token_dict.items():
        ending = ' ' if key in ['\n', '(', '"'] else ''
        tv_script = tv_script.replace(' ' + token.lower(), key)
    tv_script = tv_script.replace('\n ', '\n')
    tv_script = tv_script.replace('( ', '(')
        
    print(tv_script)
moe_szyslak: marge, i loves youse. will youse be mines?
lloyd: so, bee fast. and i can't believe him. you know, ya, ma'am and money something. no more down her and thirty same while we'll an arse to be grace and make me to survive a guy.
homer_simpson:(slyly) i want to?
moe_szyslak: yeah? oh, thanks, moe!
marge_simpson:(tipsy) keep the old woman. to the matter with this jar.
duffman: homer, i'd get back to the doug: conference dollars.
homer_simpson: i was so everybody we're a guy. but i know, here. fifty name. all uh-huh...
lenny_leonard:(upset) hey sees moe, grampa! yeah.
homer_simpson: it was like our hollowed-out not in the bill.
moe_szyslak: six of it.
moe_szyslak: aw, in that was, your beloved of make me from me too.
moe_szyslak: heh! he floated up banks, liver machine.
moe_szyslak: hey, why did i go. and me

The TV Script is Nonsensical

It's ok if the TV script doesn't make any sense. We trained on less than a megabyte of text. In order to get good results, you'll have to use a smaller vocabulary or get more data. Luckly there's more data! As we mentioned in the begging of this project, this is a subset of another dataset. We didn't have you train on all the data, because that would take too long. However, you are free to train your neural network on all the data. After you complete the project, of course.

In [ ]:
 

No comments:

Post a Comment