In [ ]:
from __future__ import print_function
import gzip
import os
import numpy as np
import random
In [ ]:
words = open("text8.bz2").read().decode("bz2").split()
print('Data size', len(words))
In [ ]:
# Step 2: Build the dictionary and replace rare words with UNK token.
import collections
vocabulary_size = 50000
def build_dataset(words):
count = [('UNK', 0)]
count.extend(collections.Counter(words).most_common(vocabulary_size - 1))
dictionary = {word: i for i, (word, _) in enumerate(count)}
data = [dictionary.get(word, 0) for word in words]
reverse_dictionary = {v: k for k, v in dictionary.iteritems()}
unk_count = len(words)-sum(w[1] for w in count)
count[0] = ('UNK', unk_count)
return data, count, dictionary, reverse_dictionary
data, count, dictionary, reverse_dictionary = build_dataset(words)
del words # Hint to reduce memory.
print('Most common words (+UNK)', count[:5])
print('Sample data', data[:10])
data_index = 0
In [ ]:
# Step 4: Function to generate a training batch for the skip-gram model.
def generate_batch(batch_size, num_skips, skip_window):
global data_index
assert batch_size % num_skips == 0
assert num_skips <= 2 * skip_window
batch = np.ndarray(shape=(batch_size), dtype=np.int32)
labels = np.ndarray(shape=(batch_size, 1), dtype=np.int32)
span = 2 * skip_window + 1 # [ skip_window target skip_window ]
buffer = collections.deque(maxlen=span)
for _ in range(span):
buffer.append(data[data_index])
data_index = (data_index + 1) % len(data)
for i in range(batch_size / num_skips):
target = skip_window # target label at the center of the buffer
targets_to_avoid = [ skip_window ]
for j in range(num_skips):
while target in targets_to_avoid:
target = random.randint(0, span - 1)
targets_to_avoid.append(target)
batch[i * num_skips + j] = buffer[skip_window]
labels[i * num_skips + j, 0] = buffer[target]
buffer.append(data[data_index])
data_index = (data_index + 1) % len(data)
return batch, labels
batch, labels = generate_batch(batch_size=8, num_skips=2, skip_window=1)
for i in range(8):
print(batch[i], '->', labels[i, 0])
print(reverse_dictionary[batch[i]], '->', reverse_dictionary[labels[i, 0]])
In [ ]:
import tensorflow as tf
import math
In [ ]:
def gpu_node(n):
# print(n.type)
if n.type == "MatMUl":
return "/gpu:0"
else:
return "/cpu:0"
In [ ]:
# Step 5: Build and train a skip-gram model.
batch_size = 128
embedding_size = 128 # Dimension of the embedding vector.
skip_window = 3 # How many words to consider left and right.
num_skips = 4 # How many times to reuse an input to generate a label.
# We pick a random validation set to sample nearest neighbors. Here we limit the
# validation samples to the words that have a low numeric ID, which by
# construction are also the most frequent.
valid_size = 16 # Random set of words to evaluate similarity on.
valid_window = 100 # Only pick dev samples in the head of the distribution.
valid_examples = np.array(random.sample(xrange(valid_window), valid_size))
num_sampled = 64 # Number of negative examples to sample.
graph = tf.Graph()
with graph.as_default():
with graph.device(gpu_node):
# Input data.
train_inputs = tf.placeholder(tf.int32, shape=[batch_size])
train_labels = tf.placeholder(tf.int32, shape=[batch_size, 1])
valid_dataset = tf.constant(valid_examples, dtype=tf.int32)
# Construct the variables.
embeddings = tf.Variable(
tf.random_uniform([vocabulary_size, embedding_size], -1.0, 1.0))
nce_weights = tf.Variable(
tf.truncated_normal([vocabulary_size, embedding_size],
stddev=1.0 / math.sqrt(embedding_size)))
nce_biases = tf.Variable(tf.zeros([vocabulary_size]))
# Look up embeddings for inputs.
embed = tf.nn.embedding_lookup(embeddings, train_inputs)
# Compute the average NCE loss for the batch.
# tf.nce_loss automatically draws a new sample of the negative labels each
# time we evaluate the loss.
loss = tf.reduce_mean(
tf.nn.nce_loss(nce_weights, nce_biases, embed, train_labels,
num_sampled, vocabulary_size))
# Construct the SGD optimizer using a learning rate of 1.0.
optimizer = tf.train.GradientDescentOptimizer(1.0).minimize(loss)
# Compute the cosine similarity between minibatch examples and all embeddings.
norm = tf.sqrt(tf.reduce_sum(tf.square(embeddings), 1, keep_dims=True))
normalized_embeddings = embeddings / norm
valid_embeddings = tf.nn.embedding_lookup(normalized_embeddings, valid_dataset)
similarity = tf.matmul(valid_embeddings, normalized_embeddings, transpose_b=True)
In [ ]:
session = tf.InteractiveSession(graph=graph)
In [ ]:
session.run(tf.initialize_all_variables())
In [ ]:
# Step 6: Begin training
num_steps = 2000001
with session.as_default():
average_loss = 0
for step in xrange(num_steps):
batch_inputs, batch_labels = generate_batch(
batch_size, num_skips, skip_window)
feed_dict = {train_inputs : batch_inputs, train_labels : batch_labels}
# We perform one update step by evaluating the optimizer op (including it
# in the list of returned values for session.run()
_, loss_val = session.run([optimizer, loss], feed_dict=feed_dict)
average_loss += loss_val
if step % 50000 == 0:
if step > 0:
average_loss = average_loss / 50000
# The average loss is an estimate of the loss over the last 2000 batches.
print("Average loss at step ", step, ": ", average_loss)
average_loss = 0
# note that this is expensive (~20% slowdown if computed every 500 steps)
if step % 500000 == 0:
sim = similarity.eval()
for i in xrange(valid_size):
valid_word = reverse_dictionary[valid_examples[i]]
top_k = 8 # number of nearest neighbors
nearest = (-sim[i, :]).argsort()[1:top_k+1]
log_str = "Nearest to %s:" % valid_word
for k in xrange(top_k):
close_word = reverse_dictionary[nearest[k]]
log_str = "%s %s," % (log_str, close_word)
print(log_str)
result_embeddings = embeddings.eval()
final_embeddings = normalized_embeddings.eval()
In [ ]:
import matplotlib
import matplotlib.pyplot as plt
%matplotlib inline
In [ ]:
def plot_with_labels(low_dim_embs, labels, filename='tsne.png'):
assert low_dim_embs.shape[0] >= len(labels), "More labels than embeddings"
plt.figure(figsize=(18, 18)) #in inches
for i, label in enumerate(labels):
x, y = low_dim_embs[i,:]
plt.scatter(x, y)
plt.annotate(label,
xy=(x, y),
xytext=(5, 2),
textcoords='offset points',
ha='right',
va='bottom')
plt.savefig(filename)
try:
from sklearn.manifold import TSNE
import matplotlib.pyplot as plt
tsne = TSNE(perplexity=30, n_components=2, init='pca', n_iter=5000)
plot_only = 500
low_dim_embs = tsne.fit_transform(final_embeddings[:plot_only,:])
labels = dictionary.keys()[:plot_only]
plot_with_labels(low_dim_embs, labels)
except ImportError:
print("Please install sklearn and matplotlib to visualize embeddings.")
In [ ]:
def find_sim(v, num=10):
#return np.dot(final_embeddings, v).argsort()[-num:][::-1]
return [reverse_dictionary[x] for x in np.dot(final_embeddings, v).argsort()[-num:][::-1]]
def w2v(w):
return result_embeddings[dictionary.get(w, 0)]
print(find_sim(w2v('four')))
In [ ]:
target = find_sim(w2v('king')-w2v('man')+w2v('woman'))
print(target)
In [ ]:
v1 = w2v('king')-w2v('man')+w2v('woman')
v2 = w2v('queen')
np.dot(v1, v2) /np.linalg.norm(v1)/np.linalg.norm(v2)
In [ ]:
v1 = w2v('king')-w2v('man')+w2v('woman')
v2 = w2v(target[0])
np.dot(v1, v2) /np.linalg.norm(v1)/np.linalg.norm(v2)
In [ ]:
np.linalg.norm(v1), np.linalg.norm(v2)
In [ ]:
find_sim(w2v('eat')-w2v('food'))
In [ ]:
with graph.as_default():
saver = tf.train.Saver()
In [ ]:
save_path = saver.save(session, "save/word2vec.ckpt")
print("Model saved in file: ", save_path)
In [ ]:
saver.restore(session, "save/word2vec.ckpt")
In [ ]: