Load pre-trained word embedding into Tensorflow PTB LSTM language model tutorial

ptb_word_lm_embed.py

# Copyright 2015 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================

"""Modifies the Tensorflow PTB LSTM model: 
https://github.com/tensorflow/models/blob/master/tutorials/rnn/ptb/ptb_word_lm.py
to load the GoogleNews word embedding



Example / benchmark for building a PTB LSTM model.

Trains the model described in:
(Zaremba, et. al.) Recurrent Neural Network Regularization
http://arxiv.org/abs/1409.2329

There are 3 supported model configurations:
===========================================
| config | epochs | train | valid  | test
===========================================
| small  | 13     | 37.99 | 121.39 | 115.91
| medium | 39     | 48.45 |  86.16 |  82.07
| large  | 55     | 37.87 |  82.62 |  78.29
The exact results may vary depending on the random initialization.

The hyperparameters used in the model:
- init_scale - the initial scale of the weights
- learning_rate - the initial value of the learning rate
- max_grad_norm - the maximum permissible norm of the gradient
- num_layers - the number of LSTM layers
- num_steps - the number of unrolled steps of LSTM
- hidden_size - the number of LSTM units
- max_epoch - the number of epochs trained with the initial learning rate
- max_max_epoch - the total number of epochs for training
- keep_prob - the probability of keeping weights in the dropout layer
- lr_decay - the decay of the learning rate for each epoch after "max_epoch"
- batch_size - the batch size

The data required for this example is in the data/ dir of the
PTB dataset from Tomas Mikolov's webpage:

$ wget http://www.fit.vutbr.cz/~imikolov/rnnlm/simple-examples.tgz
$ tar xvf simple-examples.tgz

To run:

$ python ptb_word_lm_embed.py --data_path=simple-examples/data/

"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import time
import os

import numpy as np
import tensorflow as tf

import reader

flags = tf.flags
logging = tf.logging

flags.DEFINE_string(
    "model", "small",
    "A type of model. Possible options are: small, medium, large.")
flags.DEFINE_string("data_path", None,
                    "Where the training/test data is stored.")
flags.DEFINE_string("save_path", None,
                    "Model output directory.")
flags.DEFINE_bool("use_fp16", False,
                  "Train using 16-bit floats instead of 32bit floats")

FLAGS = flags.FLAGS


def data_type():
    return tf.float16 if FLAGS.use_fp16 else tf.float32


class PTBInput(object):
    """The input data."""

    def __init__(self, config, data, name=None):
        self.batch_size = batch_size = config.batch_size
        self.num_steps = num_steps = config.num_steps
        self.epoch_size = ((len(data) // batch_size) - 1) // num_steps
        self.input_data, self.targets = reader.ptb_producer(
            data, batch_size, num_steps, name=name)


class PTBModel(object):
    """The PTB model."""

    def __init__(self, is_training, config, input_):
        self._input = input_

        batch_size = input_.batch_size
        num_steps = input_.num_steps
        size = config.hidden_size
        vocab_size = config.vocab_size

        # Slightly better results can be obtained with forget gate biases
        # initialized to 1 but the hyperparameters of the model would need to be
        # different than reported in the paper.
        def lstm_cell():
            return tf.contrib.rnn.BasicLSTMCell(
                size, forget_bias=0.0, state_is_tuple=True)
        attn_cell = lstm_cell
        if is_training and config.keep_prob < 1:
            def attn_cell():
                return tf.contrib.rnn.DropoutWrapper(
                    lstm_cell(), output_keep_prob=config.keep_prob)
        cell = tf.contrib.rnn.MultiRNNCell(
            [attn_cell() for _ in range(config.num_layers)], state_is_tuple=True)

        self._initial_state = cell.zero_state(batch_size, data_type())

        with tf.device("/cpu:0"):
            self.embedding = tf.get_variable(
                "embedding", [vocab_size, size], dtype=data_type(), trainable=False)
            inputs = tf.nn.embedding_lookup(self.embedding, input_.input_data)
        if is_training and config.keep_prob < 1:
            inputs = tf.nn.dropout(inputs, config.keep_prob)

        # Simplified version of models/tutorials/rnn/rnn.py's rnn().
        # This builds an unrolled LSTM for tutorial purposes only.
        # In general, use the rnn() or state_saving_rnn() from rnn.py.
        #
        # The alternative version of the code below is:
        #
        # inputs = tf.unstack(inputs, num=num_steps, axis=1)
        # outputs, state = tf.nn.rnn(cell, inputs,
        #                            initial_state=self._initial_state)
        outputs = []
        state = self._initial_state
        with tf.variable_scope("RNN"):
            for time_step in range(num_steps):
                if time_step > 0: tf.get_variable_scope().reuse_variables()
                (cell_output, state) = cell(inputs[:, time_step, :], state)
                outputs.append(cell_output)

        output = tf.reshape(tf.concat(axis=1, values=outputs), [-1, size])
        softmax_w = tf.get_variable(
            "softmax_w", [size, vocab_size], dtype=data_type())
        softmax_b = tf.get_variable("softmax_b", [vocab_size], dtype=data_type())
        logits = tf.matmul(output, softmax_w) + softmax_b
        loss = tf.contrib.legacy_seq2seq.sequence_loss_by_example(
            [logits],
            [tf.reshape(input_.targets, [-1])],
            [tf.ones([batch_size * num_steps], dtype=data_type())])
        self._cost = cost = tf.reduce_sum(loss) / batch_size
        self._final_state = state

        if not is_training:
            return

        self._lr = tf.Variable(0.0, trainable=False)
        tvars = tf.trainable_variables()
        grads, _ = tf.clip_by_global_norm(tf.gradients(cost, tvars),
                                          config.max_grad_norm)
        optimizer = tf.train.GradientDescentOptimizer(self._lr)
        self._train_op = optimizer.apply_gradients(
            zip(grads, tvars),
            global_step=tf.contrib.framework.get_or_create_global_step())

        self._new_lr = tf.placeholder(
            tf.float32, shape=[], name="new_learning_rate")
        self._lr_update = tf.assign(self._lr, self._new_lr)

    def assign_lr(self, session, lr_value):
        session.run(self._lr_update, feed_dict={self._new_lr: lr_value})

    @property
    def input(self):
        return self._input

    @property
    def initial_state(self):
        return self._initial_state

    @property
    def cost(self):
        return self._cost

    @property
    def final_state(self):
        return self._final_state

    @property
    def lr(self):
        return self._lr

    @property
    def train_op(self):
        return self._train_op


class SmallConfig(object):
    """Small config."""
    init_scale = 0.1
    learning_rate = 1.0
    max_grad_norm = 5
    num_layers = 2
    num_steps = 20
    hidden_size = 300
    max_epoch = 4
    max_max_epoch = 13
    keep_prob = 1.0
    lr_decay = 0.5
    batch_size = 20
    vocab_size = 10000


class MediumConfig(object):
    """Medium config."""
    init_scale = 0.05
    learning_rate = 1.0
    max_grad_norm = 5
    num_layers = 2
    num_steps = 35
    hidden_size = 650
    max_epoch = 6
    max_max_epoch = 39
    keep_prob = 0.5
    lr_decay = 0.8
    batch_size = 20
    vocab_size = 10000


class LargeConfig(object):
    """Large config."""
    init_scale = 0.04
    learning_rate = 1.0
    max_grad_norm = 10
    num_layers = 2
    num_steps = 35
    hidden_size = 1500
    max_epoch = 14
    max_max_epoch = 55
    keep_prob = 0.35
    lr_decay = 1 / 1.15
    batch_size = 20
    vocab_size = 10000


class TestConfig(object):
    """Tiny config, for testing."""
    init_scale = 0.1
    learning_rate = 1.0
    max_grad_norm = 1
    num_layers = 1
    num_steps = 2
    hidden_size = 2
    max_epoch = 1
    max_max_epoch = 1
    keep_prob = 1.0
    lr_decay = 0.5
    batch_size = 20
    vocab_size = 10000


def run_epoch(session, model, eval_op=None, verbose=False):
    """Runs the model on the given data."""
    start_time = time.time()
    costs = 0.0
    iters = 0
    state = session.run(model.initial_state)

    fetches = {
        "cost": model.cost,
        "final_state": model.final_state,
    }
    if eval_op is not None:
        fetches["eval_op"] = eval_op

    for step in range(model.input.epoch_size):
        feed_dict = {}
        for i, (c, h) in enumerate(model.initial_state):
            feed_dict[c] = state[i].c
            feed_dict[h] = state[i].h

        vals = session.run(fetches, feed_dict)
        cost = vals["cost"]
        state = vals["final_state"]

        costs += cost
        iters += model.input.num_steps

        if verbose and step % (model.input.epoch_size // 10) == 10:
            print("%.3f perplexity: %.3f speed: %.0f wps" %
                  (step * 1.0 / model.input.epoch_size, np.exp(costs / iters),
                   iters * model.input.batch_size / (time.time() - start_time)))

    return np.exp(costs / iters)

def loadEmbedding(word_to_id):
    """ Initialize embeddings with pre-trained word2vec vectors
        Will modify the embedding weights of the current loaded model
        Uses the GoogleNews pre-trained values (path hardcoded)
        """
    # Load the pre-trained word2vec data
    with open('GoogleNews-vectors-negative300.bin', "rb", 0) as f:
        header = f.readline()
        vocab_size, vector_size = map(int, header.split())
        binary_len = np.dtype('float32').itemsize * vector_size
        initW = np.random.uniform(-0.25,0.25,(len(word_to_id), vector_size))
        for line in range(vocab_size):
            word = []
            while True:
                ch = f.read(1)
                if ch == b' ':
                    word = b''.join(word).decode('utf-8')
                    break
                if ch != b'\n':
                    word.append(ch)
            if word in word_to_id:
                initW[word_to_id[word]] = np.fromstring(f.read(binary_len), dtype='float32')
            else:
                f.read(binary_len)

    # PCA Decomposition to reduce word2vec dimensionality
    #     U, s, Vt = np.linalg.svd(initW, full_matrices=False)
    #     S = np.zeros((vector_size, vector_size), dtype=complex)
    #     S[:vector_size, :vector_size] = np.diag(s)
    #     initW = np.dot(U[:, :self.args.embeddingSize], S[:self.args.embeddingSize, :self.args.embeddingSize])

    return initW

def get_config():
    if FLAGS.model == "small":
        return SmallConfig()
    elif FLAGS.model == "medium":
        return MediumConfig()
    elif FLAGS.model == "large":
        return LargeConfig()
    elif FLAGS.model == "test":
        return TestConfig()
    else:
        raise ValueError("Invalid model: %s", FLAGS.model)


def main(_):
    if not FLAGS.data_path:
        raise ValueError("Must set --data_path to PTB data directory")

    raw_data = reader.ptb_raw_data(FLAGS.data_path)
    train_data, valid_data, test_data, _, word_to_id = raw_data

    config = get_config()
    eval_config = get_config()
    eval_config.batch_size = 1
    eval_config.num_steps = 1

    with tf.Graph().as_default():
        initializer = tf.random_uniform_initializer(-config.init_scale,
                                                    config.init_scale)
        with tf.name_scope("Train"):
            train_input = PTBInput(config=config, data=train_data, name="TrainInput")
            with tf.variable_scope("Model", reuse=None, initializer=initializer):
                m = PTBModel(is_training=True, config=config, input_=train_input)
            word2vec = loadEmbedding(word_to_id)
            assign_embedding = tf.assign(m.embedding, word2vec)
            tf.summary.scalar("Training Loss", m.cost)
            tf.summary.scalar("Learning Rate", m.lr)

        with tf.name_scope("Valid"):
            valid_input = PTBInput(config=config, data=valid_data, name="ValidInput")
            with tf.variable_scope("Model", reuse=True, initializer=initializer):
                mvalid = PTBModel(is_training=False, config=config, input_=valid_input)
            tf.summary.scalar("Validation Loss", mvalid.cost)

        with tf.name_scope("Test"):
            test_input = PTBInput(config=eval_config, data=test_data, name="TestInput")
            with tf.variable_scope("Model", reuse=True, initializer=initializer):
                mtest = PTBModel(is_training=False, config=eval_config,
                                 input_=test_input)

        sv = tf.train.Supervisor(logdir=FLAGS.save_path)
        with sv.managed_session() as session:
            session.run(assign_embedding)
            for i in range(config.max_max_epoch):
                lr_decay = config.lr_decay ** max(i + 1 - config.max_epoch, 0.0)
                m.assign_lr(session, config.learning_rate * lr_decay)

                print("Epoch: %d Learning rate: %.3f" % (i + 1, session.run(m.lr)))
                train_perplexity = run_epoch(session, m, eval_op=m.train_op,
                                             verbose=True)
                print("Epoch: %d Train Perplexity: %.3f" % (i + 1, train_perplexity))
                valid_perplexity = run_epoch(session, mvalid)
                print("Epoch: %d Valid Perplexity: %.3f" % (i + 1, valid_perplexity))

                test_perplexity = run_epoch(session, mtest)
                print("Test Perplexity: %.3f" % test_perplexity)

                if FLAGS.save_path:
                    print("Saving model to %s." % FLAGS.save_path)
                    sv.saver.save(session, FLAGS.save_path, global_step=sv.global_step)


if __name__ == "__main__":
    tf.app.run()

reader.py

# Copyright 2015 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================


"""Utilities for parsing PTB text files."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import collections
import os

import tensorflow as tf


def _read_words(filename):
  with tf.gfile.GFile(filename, "r") as f:
    return f.read().decode("utf-8").replace("\n", "<eos>").split()


def _build_vocab(filename):
  data = _read_words(filename)

  counter = collections.Counter(data)
  count_pairs = sorted(counter.items(), key=lambda x: (-x[1], x[0]))

  words, _ = list(zip(*count_pairs))
  word_to_id = dict(zip(words, range(len(words))))

  return word_to_id


def _file_to_word_ids(filename, word_to_id):
  data = _read_words(filename)
  return [word_to_id[word] for word in data if word in word_to_id]


def ptb_raw_data(data_path=None):
  """Load PTB raw data from data directory "data_path".

  Reads PTB text files, converts strings to integer ids,
  and performs mini-batching of the inputs.

  The PTB dataset comes from Tomas Mikolov's webpage:

  http://www.fit.vutbr.cz/~imikolov/rnnlm/simple-examples.tgz

  Args:
    data_path: string path to the directory where simple-examples.tgz has
      been extracted.

  Returns:
    tuple (train_data, valid_data, test_data, vocabulary)
    where each of the data objects can be passed to PTBIterator.
  """

  train_path = os.path.join(data_path, "ptb.train.txt")
  valid_path = os.path.join(data_path, "ptb.valid.txt")
  test_path = os.path.join(data_path, "ptb.test.txt")

  word_to_id = _build_vocab(train_path)
  train_data = _file_to_word_ids(train_path, word_to_id)
  valid_data = _file_to_word_ids(valid_path, word_to_id)
  test_data = _file_to_word_ids(test_path, word_to_id)
  vocabulary = len(word_to_id)
  return train_data, valid_data, test_data, vocabulary, word_to_id


def ptb_producer(raw_data, batch_size, num_steps, name=None):
  """Iterate on the raw PTB data.

  This chunks up raw_data into batches of examples and returns Tensors that
  are drawn from these batches.

  Args:
    raw_data: one of the raw data outputs from ptb_raw_data.
    batch_size: int, the batch size.
    num_steps: int, the number of unrolls.
    name: the name of this operation (optional).

  Returns:
    A pair of Tensors, each shaped [batch_size, num_steps]. The second element
    of the tuple is the same data time-shifted to the right by one.

  Raises:
    tf.errors.InvalidArgumentError: if batch_size or num_steps are too high.
  """
  with tf.name_scope(name, "PTBProducer", [raw_data, batch_size, num_steps]):
    raw_data = tf.convert_to_tensor(raw_data, name="raw_data", dtype=tf.int32)

    data_len = tf.size(raw_data)
    batch_len = data_len // batch_size
    data = tf.reshape(raw_data[0 : batch_size * batch_len],
                      [batch_size, batch_len])

    epoch_size = (batch_len - 1) // num_steps
    assertion = tf.assert_positive(
        epoch_size,
        message="epoch_size == 0, decrease batch_size or num_steps")
    with tf.control_dependencies([assertion]):
      epoch_size = tf.identity(epoch_size, name="epoch_size")

    i = tf.train.range_input_producer(epoch_size, shuffle=False).dequeue()
    x = tf.strided_slice(data, [0, i * num_steps],
                         [batch_size, (i + 1) * num_steps])
    x.set_shape([batch_size, num_steps])
    y = tf.strided_slice(data, [0, i * num_steps + 1],
                         [batch_size, (i + 1) * num_steps + 1])
    y.set_shape([batch_size, num_steps])
    return x, y