Adapted from https://github.com/keras-team/keras/blob/master/examples/addition_rnn.py
In [1]:
!pip install -q tf-nightly-gpu-2.0-preview
In [2]:
import tensorflow as tf
print(tf.__version__)
In [0]:
class CharacterTable(object):
"""Given a set of characters:
+ Encode them to a one hot integer representation
+ Decode the one hot integer representation to their character output
+ Decode a vector of probabilities to their character output
"""
def __init__(self, chars):
"""Initialize character table.
# Arguments
chars: Characters that can appear in the input.
"""
self.chars = sorted(set(chars))
self.char_indices = dict((c, i) for i, c in enumerate(self.chars))
self.indices_char = dict((i, c) for i, c in enumerate(self.chars))
def encode(self, C, num_rows):
"""One hot encode given string C.
# Arguments
num_rows: Number of rows in the returned one hot encoding. This is
used to keep the # of rows for each data the same.
"""
x = np.zeros((num_rows, len(self.chars)))
for i, c in enumerate(C):
x[i, self.char_indices[c]] = 1
return x
def decode(self, x, calc_argmax=True):
if calc_argmax:
x = x.argmax(axis=-1)
return ''.join(self.indices_char[x] for x in x)
In [0]:
class colors:
ok = '\033[92m'
fail = '\033[91m'
close = '\033[0m'
In [6]:
import numpy as np
# Parameters for the model and dataset.
TRAINING_SIZE = 50000
DIGITS = 3
# REVERSE = True
REVERSE = False
# Maximum length of input is 'int + int' (e.g., '345+678'). Maximum length of
# int is DIGITS.
MAXLEN = DIGITS + 1 + DIGITS
# All the numbers, plus sign and space for padding.
chars = '0123456789+ '
ctable = CharacterTable(chars)
questions = []
expected = []
seen = set()
print('Generating data...')
while len(questions) < TRAINING_SIZE:
f = lambda: int(''.join(np.random.choice(list('0123456789'))
for i in range(np.random.randint(1, DIGITS + 1))))
a, b = f(), f()
# Skip any addition questions we've already seen
# Also skip any such that x+Y == Y+x (hence the sorting).
key = tuple(sorted((a, b)))
if key in seen:
continue
seen.add(key)
# Pad the data with spaces such that it is always MAXLEN.
q = '{}+{}'.format(a, b)
query = q + ' ' * (MAXLEN - len(q))
ans = str(a + b)
# Answers can be of maximum size DIGITS + 1.
ans += ' ' * (DIGITS + 1 - len(ans))
if REVERSE:
# Reverse the query, e.g., '12+345 ' becomes ' 543+21'. (Note the
# space used for padding.)
query = query[::-1]
questions.append(query)
expected.append(ans)
print('Total addition questions:', len(questions))
In [13]:
questions[0]
Out[13]:
In [14]:
expected[0]
Out[14]:
In [8]:
print('Vectorization...')
x = np.zeros((len(questions), MAXLEN, len(chars)), dtype=np.bool)
y = np.zeros((len(questions), DIGITS + 1, len(chars)), dtype=np.bool)
for i, sentence in enumerate(questions):
x[i] = ctable.encode(sentence, MAXLEN)
for i, sentence in enumerate(expected):
y[i] = ctable.encode(sentence, DIGITS + 1)
In [9]:
len(x[0])
Out[9]:
In [10]:
len(questions[0])
Out[10]:
In [12]:
x[0]
Out[12]:
In [15]:
y[0]
Out[15]:
In [0]:
# Shuffle (x, y) in unison as the later parts of x will almost all be larger
# digits.
indices = np.arange(len(y))
np.random.shuffle(indices)
x = x[indices]
y = y[indices]
In [17]:
# Explicitly set apart 10% for validation data that we never train over.
split_at = len(x) - len(x) // 10
(x_train, x_val) = x[:split_at], x[split_at:]
(y_train, y_val) = y[:split_at], y[split_at:]
print('Training Data:')
print(x_train.shape)
print(y_train.shape)
print('Validation Data:')
print(x_val.shape)
print(y_val.shape)
In [18]:
# input shape: 7 digits, each being 0-9, + or space (12 possibilities)
MAXLEN, len(chars)
Out[18]:
In [23]:
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import LSTM, GRU, SimpleRNN, Dense, RepeatVector
# Try replacing LSTM, GRU, or SimpleRNN.
# RNN = LSTM
RNN = SimpleRNN # should be enough since we do not have long sequences and only local dependencies
# RNN = GRU
HIDDEN_SIZE = 128
BATCH_SIZE = 128
model = Sequential()
# encoder
model.add(RNN(units=HIDDEN_SIZE, input_shape=(MAXLEN, len(chars))))
# latent space
encoding_dim = 32
model.add(Dense(units=encoding_dim, activation='relu', name="encoder"))
# decoder: have 4 temporal outputs one for each of the digits of the results
model.add(RepeatVector(DIGITS + 1))
# return_sequences=True tells it to keep all 4 temporal outputs, not only the final one (we need all four digits for the results)
model.add(RNN(units=HIDDEN_SIZE, return_sequences=True))
model.add(Dense(name='classifier', units=len(chars), activation='softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
model.summary()
In [24]:
%%time
# Train the model each generation and show predictions against the validation
# dataset.
merged_losses = {
"loss": [],
"val_loss": [],
"accuracy": [],
"val_accuracy": [],
}
for iteration in range(1, 50):
print()
print('-' * 50)
print('Iteration', iteration)
iteration_history = model.fit(x_train, y_train,
batch_size=BATCH_SIZE,
epochs=1,
validation_data=(x_val, y_val))
merged_losses["loss"].append(iteration_history.history["loss"])
merged_losses["val_loss"].append(iteration_history.history["val_loss"])
merged_losses["accuracy"].append(iteration_history.history["accuracy"])
merged_losses["val_accuracy"].append(iteration_history.history["val_accuracy"])
# Select 10 samples from the validation set at random so we can visualize
# errors.
for i in range(10):
ind = np.random.randint(0, len(x_val))
rowx, rowy = x_val[np.array([ind])], y_val[np.array([ind])]
preds = model.predict_classes(rowx, verbose=0)
q = ctable.decode(rowx[0])
correct = ctable.decode(rowy[0])
guess = ctable.decode(preds[0], calc_argmax=False)
print('Q', q[::-1] if REVERSE else q, end=' ')
print('T', correct, end=' ')
if correct == guess:
print(colors.ok + '☑' + colors.close, end=' ')
else:
print(colors.fail + '☒' + colors.close, end=' ')
print(guess)
In [25]:
import matplotlib.pyplot as plt
plt.ylabel('loss')
plt.xlabel('epoch')
plt.yscale('log')
plt.plot(merged_losses['loss'])
plt.plot(merged_losses['val_loss'])
plt.legend(['loss', 'validation loss'])
Out[25]:
In [27]:
plt.ylabel('accuracy')
plt.xlabel('epoch')
# plt.yscale('log')
plt.plot(merged_losses['accuracy'])
plt.plot(merged_losses['val_accuracy'])
plt.legend(['accuracy', 'validation accuracy'])
Out[27]:
In [0]: