TV Script Generation¶
In this project, you'll generate your own Simpsons TV scripts using RNNs. You'll be using part of the Simpsons dataset of scripts from 27 seasons. The Neural Network you'll build will generate a new TV script for a scene at Moe's Tavern.
Get the Data¶
The data is already provided for you. You'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..
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
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:]
Explore the Data¶
Play around with view_sentence_range to view different parts of the data.
view_sentence_range = (0, 10)
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
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]]))
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, you first need to transform the words to ids. In this function, 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
Return these dictionaries in the following tuple (vocab_to_int, int_to_vocab)
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 = sorted(set(text))
vocab_to_int = {word: index for index, word in enumerate(vocab)}
int_to_vocab = dict(enumerate(vocab))
return vocab_to_int, int_to_vocab
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_create_lookup_tables(create_lookup_tables)
Optional - Text analysis¶
from collections import Counter
words = sentences.copy()
cleaned_words = [row.translate({ord(i):'' for i in '\'"():,.!?@#$\n'}) for row in words]
print("Original text")
print(words[:10])
print("\nCleaned text:")
print(cleaned_words[:10])
word_list = []
for row in cleaned_words:
for word in row:
word_list.append(word)
word_list = ''.join(word_list)
word_list = word_list.split()
total_count = len(word_list)
word_counter = Counter(word_list)
frequencies = [count/total_count for word, count in word_counter.items()]
freq_indices = np.argsort(-np.array(frequencies))
vocab_to_int, int_to_vocab = create_lookup_tables(word_list)
print("\nOverview of most common words:")
count = 0
for index in freq_indices:
word = word_list[index]
print("{} {} {:0.1f}%".format(word, index, frequencies[index]*100))
count += 1
if count==15:
break
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||".
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 symbol to token conversion
symbol_to_token = {'.' : '||Period||',
',' : '||Comma||',
'"' : '||QuotationMark||',
';' : '||Semicolon||',
'!' : '||ExclamationMark||',
'?': '||QuestionMark||',
'(': '||LeftParentheses||',
')': '||RightParentheses||',
'--': '||Dash||',
'\n' : '||Return||'}
# token_to_symbol = {token: key for value, key in symbol_to_token}
return symbol_to_token
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_tokenize(token_lookup)
Tests¶
lookup = token_lookup()
for token in ['.',',', '"',';','!','?','(', ')', '--', '\n']:
print("{} --> {}".format('(return)' if token=='\n' else token, lookup[token]))
Preprocess all the data and save it¶
Running the code cell below will preprocess all the data and save it to file.
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
# 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.
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper
import numpy as np
import problem_unittests as tests
int_text, vocab_to_int, int_to_vocab, token_dict = helper.load_preprocess()
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
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()))
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
nameparameter. - Targets placeholder
- Learning Rate placeholder
Return the placeholders in the following tuple (Input, Targets, LearningRate)
def get_inputs():
"""
Create TF Placeholders for input, targets, and learning rate.
:return: Tuple (input, targets, learning rate)
"""
# TODO: Implement Function
# Input is a list of sequences of words
input = tf.placeholder(tf.int32, [None, None], name='input')
# Training target is a shifted sequence, see batch creation function below
targets = tf.placeholder(tf.int32, [None, None], name='targets')
# Learning rate, will later be defined below in the hyperparameters section
learning_rate = tf.placeholder(tf.float32, name = 'learning_rate')
return input, targets, learning_rate
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_get_inputs(get_inputs)
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()
- Apply the name "initial_state" to the initial state using
Return the cell and initial state in the following tuple (Cell, InitialState)
# Constant hyper parameters
# Chance by which a neuron is kept active. 70% led to very satisfying results
training_dropout_keep = 0.7
# Count of RNN layers
rnn_layers = 2
def create_cell(rnn_size):
lstm = tf.contrib.rnn.BasicLSTMCell(rnn_size)
drop = tf.contrib.rnn.DropoutWrapper(lstm, output_keep_prob=training_dropout_keep)
return drop
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
cell = tf.contrib.rnn.MultiRNNCell([create_cell(rnn_size) for i in range(rnn_layers)])
zero_state = cell.zero_state(batch_size, tf.float32)
initial_state = tf.identity(zero_state, name="initial_state")
return cell, initial_state
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_get_init_cell(get_init_cell)
Word Embedding¶
Apply embedding to input_data using TensorFlow. Return the embedded sequence.
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(initial_value=tf.random_uniform(shape=[vocab_size, embed_dim], minval=-1.0, maxval=1.0))
embed = tf.nn.embedding_lookup(embedding, input_data)
return embed
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_get_embed(get_embed)
Build RNN¶
You created a RNN Cell in the get_init_cell() function. Time to use the cell to create a RNN.
- Build the RNN using the
tf.nn.dynamic_rnn()- Apply the name "final_state" to the final state using
tf.identity()
- Apply the name "final_state" to the final state using
Return the outputs and final_state state in the following tuple (Outputs, FinalState)
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
# Construct dynamic neural network
outputs, state = tf.nn.dynamic_rnn(cell, inputs, dtype=tf.float32)
# Setup initial state
state = tf.identity(state, name='final_state')
return outputs, state
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_build_rnn(build_rnn)
Build the Neural Network¶
Apply the functions you implemented above to:
- Apply embedding to
input_datausing yourget_embed(input_data, vocab_size, embed_dim)function. - Build RNN using
celland yourbuild_rnn(cell, inputs)function. - Apply a fully connected layer with a linear activation and
vocab_sizeas the number of outputs.
Return the logits and final state in the following tuple (Logits, FinalState)
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
# Setup RNN
inputs = get_embed(input_data, vocab_size, embed_dim)
outputs, final_state = build_rnn(cell, inputs)
# Setup fully connected layer
logits = tf.contrib.layers.fully_connected(outputs, vocab_size, activation_fn=None)
return logits, final_state
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_build_nn(build_nn)
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, 16, 17, 18, 19, 20], 3, 2) would return a Numpy array of the following:
[
# First Batch
[
# Batch of Input
[[ 1 2], [ 7 8], [13 14]]
# Batch of targets
[[ 2 3], [ 8 9], [14 15]]
]
# Second Batch
[
# Batch of Input
[[ 3 4], [ 9 10], [15 16]]
# Batch of targets
[[ 4 5], [10 11], [16 17]]
]
# Third Batch
[
# Batch of Input
[[ 5 6], [11 12], [17 18]]
# Batch of targets
[[ 6 7], [12 13], [18 1]]
]
]Notice that the last target value in the last batch is the first input value of the first batch. In this case, 1. This is a common technique used when creating sequence batches, although it is rather unintuitive.
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
"""
full_batch_size = batch_size*seq_length # effetive size per batch
full_batch_count = len(int_text)//full_batch_size # count of possible batches
column_size = full_batch_count*seq_length # size/offset caused by column eg 6 = 3*2 in the example above
batches = []
max_ind = full_batch_count*full_batch_size
# for all full batches
for fbi in range(full_batch_count):
batch = []
input_data = []
output_data = []
full_batch_off = fbi*seq_length
for bi in range(batch_size):
batch_off = bi*column_size + full_batch_off # eg 6 for the second columm
input_data.append(np.array([int_text[batch_off+seq] for seq in range(seq_length)]))
output_data.append(np.array([int_text[(batch_off+seq+1)%max_ind] for seq in range(seq_length)]))
batch.append(np.array(input_data))
batch.append(np.array(output_data))
batches.append(np.array(batch))
return np.array(batches)
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_get_batches(get_batches)
Optional - Verification¶
original = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20]
print("{} \n\n {}".format(original,get_batches(original, 3, 2)))
Neural Network Training¶
Hyperparameters¶
Tune the following parameters:
- Set
num_epochsto the number of epochs. - Set
batch_sizeto the batch size. - Set
rnn_sizeto the size of the RNNs. - Set
embed_dimto the size of the embedding. - Set
seq_lengthto the length of sequence. - Set
learning_rateto the learning rate. - Set
show_every_n_batchesto the number of batches the neural network should print progress.
# Number of Epochs
num_epochs = 100
# Batch Size
batch_size = 128
# RNN Size
rnn_size = 1024
# Embedding Dimension Size
embed_dim = 512
# Sequence Length
seq_length = 16
# Learning Rate
learning_rate = 0.001
# Show stats for every n number of epochs
show_every_n_epochs = 5
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
save_dir = './save'
Build the Graph¶
Build the graph using the neural network you implemented.
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
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 if grad is not None]
train_op = optimizer.apply_gradients(capped_gradients)
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
batches = get_batches(int_text, batch_size, seq_length)
loss_rate = []
epoch_index = []
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 batch_i==0:
loss_rate.append(train_loss)
epoch_index.append(epoch_i)
if epoch_i%show_every_n_epochs==0 or epoch_i==num_epochs-1:
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')
Optional Visualization of training loss¶
%matplotlib inline
from matplotlib import pyplot as plt
plt.plot(epoch_index, loss_rate)
plt.xlabel("Epochs")
plt.title("Training loss curve")
plt.ylabel("Training loss")
plt.show()
Save Parameters¶
Save seq_length and save_dir for generating a new TV script.
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
# Save parameters for checkpoint
helper.save_params((seq_length, save_dir))
Checkpoint¶
"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
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)
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
inputs = loaded_graph.get_tensor_by_name('input:0')
initial_state = loaded_graph.get_tensor_by_name('initial_state:0')
final_state = loaded_graph.get_tensor_by_name('final_state:0')
probs = loaded_graph.get_tensor_by_name('probs:0')
return inputs, initial_state, final_state, probs
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_get_tensors(get_tensors)
Choose Word¶
Implement the pick_word() function to select the next word using probabilities.
import random
def pick_word(probabilities, int_to_vocab, randomness = 1):
"""
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
max_top = randomness
max_words = max_top if len(probabilities)>max_top else len(probabilities)
# choose of a range of top words
top_x = np.argpartition(probabilities,-max_words)[-max_words:]
index = random.randint(0,max_words-1)
# select final word of pool
eff_index = top_x[index]
# print(max_words)
# print(len(probabilities))
# print(top_x)
# print(index)
return int_to_vocab[eff_index]
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_pick_word(pick_word)
Optional - Test function¶
# Following test should choose the banana
words = "Hello World Take The Banana And Nothing Else!"
weightings = [0,1,2,3,4,3,2,1]
print("Original text and word weightings: {} - {}".format(words, weightings))
print("\nSplit words:")
words = words.split()
print(words)
print("\nReconversion dictionary:")
word_dict = dict(enumerate(words))
print(word_dict)
print("\nHopefully a banana:")
print(pick_word(weightings, word_dict, randomness=1))
Generate TV Script¶
This will generate the TV script for you. Set gen_length to the length of TV script you want to generate.
gen_length = 200
# homer_simpson, moe_szyslak, or Barney_Gumble
prime_word = 'moe_szyslak'
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
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)
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.
Submitting This Project¶
When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_tv_script_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.