README.md 24.6 KB
Newer Older
C
choijulie 已提交
1
# Word2Vec
Y
Yu Yang 已提交
2

3
This is intended as a reference tutorial. The source code of this tutorial is located at [book/word2vec](https://github.com/PaddlePaddle/book/tree/develop/04.word2vec).
Z
zhangruiqing01 已提交
4

L
Luo Tao 已提交
5
For instructions on getting started with PaddlePaddle, see [PaddlePaddle installation guide](https://github.com/PaddlePaddle/book/blob/develop/README.md#running-the-book).
L
Luo Tao 已提交
6

C
choijulie 已提交
7
## Background Introduction
Z
zhangruiqing01 已提交
8

9
This section introduces the concept of **word embeddings**, which are vector representations of words. Word embeddings is a popular technique used in natural language processing to support applications such as search engines, advertising systems, and recommendation systems.
Z
zhangruiqing01 已提交
10

C
choijulie 已提交
11
### One-Hot Vectors
Z
zhangruiqing01 已提交
12

13
Building these applications requires us to quantify the similarity between two words or paragraphs. This calls for a new representation of all the words to make them more suitable for computation. An obvious way to achieve this is through the vector space model, where every word is represented as an **one-hot vector**.
Z
zhangruiqing01 已提交
14

C
choijulie 已提交
15
For each word, its vector representation has the corresponding entry in the vector as 1, and all other entries as 0. The lengths of one-hot vectors match the size of the dictionary. Each entry of a vector corresponds to the presence (or absence) of a word in the dictionary.
Z
zhangruiqing01 已提交
16

17
One-hot vectors are intuitive, yet they have limited usefulness. Take the example of an Internet advertising system: Suppose a customer enters the query "Mother's Day", while an ad bids for the keyword "carnations". Because the one-hot vectors of these two words are perpendicular, the metric distance (either Euclidean or cosine similarity) between them would indicate  little relevance. However, *we* know that these two queries are connected semantically, since people often gift their mothers bundles of carnation flowers on Mother's Day. This discrepancy is due to the low information capacity in each vector. That is, comparing the vector representations of two words does not assess their relevance sufficiently. To calculate their similarity accurately, we need more information, which could be learned from large amounts of data through machine learning methods.
Z
zhangruiqing01 已提交
18

C
choijulie 已提交
19 20
Like many machine learning models, word embeddings can represent knowledge in various ways. Another model may project an one-hot vector to an embedding vector of lower dimension e.g. $embedding(mother's day) = [0.3, 4.2, -1.5, ...], embedding(carnations) = [0.2, 5.6, -2.3, ...]$. Mapping one-hot vectors onto an embedded vector space has the potential to bring the embedding vectors of similar words (either semantically or usage-wise) closer to each other, so that the cosine similarity between the corresponding vectors for words like "Mother's Day" and "carnations" are no longer zero.

M
Mimee 已提交
21
A word embedding model could be a probabilistic model, a co-occurrence matrix model, or a neural network. Before people started using neural networks to generate word embedding, the traditional method was to calculate a co-occurrence matrix $X$ of words. Here, $X$ is a $|V| \times |V|$ matrix, where $X_{ij}$ represents the co-occurrence times of the $i$th and $j$th words in the vocabulary `V` within all corpus, and $|V|$ is the size of the vocabulary. By performing matrix decomposition on $X$ e.g. Singular Value Decomposition \[[5](#references)\]
Z
zhangruiqing01 已提交
22

C
choijulie 已提交
23
$$X = USV^T$$
Z
zhangruiqing01 已提交
24

C
choijulie 已提交
25
the resulting $U$ can be seen as the word embedding of all the words.
Z
zhangruiqing01 已提交
26

C
choijulie 已提交
27 28 29 30
However, this method suffers from many drawbacks:
1) Since many pairs of words don't co-occur, the co-occurrence matrix is sparse. To achieve good performance of matrix factorization, further treatment on word frequency is needed;
2) The matrix is large, frequently on the order of $10^6*10^6$;
3) We need to manually filter out stop words (like "although", "a", ...), otherwise these frequent words will affect the performance of matrix factorization.
Z
zhangruiqing01 已提交
31

C
choijulie 已提交
32
The neural network based model does not require storing huge hash tables of statistics on all of the corpus. It obtains the word embedding by learning from semantic information, hence could avoid the aforementioned problems in the traditional method. In this chapter, we will introduce the details of neural network word embedding model and how to train such model in PaddlePaddle.
Z
zhangruiqing01 已提交
33

C
choijulie 已提交
34
## Results Demonstration
Z
zhangruiqing01 已提交
35

M
Mimee 已提交
36
In this section, we use the $t-$SNE\[[4](#references)\] data visualization algorithm to draw the word embedding vectors after projecting them onto a two-dimensional space (see figure below). From the figure we can see that the semantically relevant words -- *a*, *the*, and *these* or *big* and *huge* -- are close to each other in the projected space, while irrelevant words -- *say* and *business* or *decision* and *japan* -- are far from each other.
Z
zhangruiqing01 已提交
37 38

<p align="center">
39
    <img src = "image/2d_similarity.png" width=400><br/>
C
choijulie 已提交
40
    Figure 1. Two dimension projection of word embeddings
Z
zhangruiqing01 已提交
41 42
</p>

C
choijulie 已提交
43 44 45
### Cosine Similarity

On the other hand, we know that the cosine similarity between two vectors falls between $[-1,1]$. Specifically, the cosine similarity is 1 when the vectors are identical, 0 when the vectors are perpendicular, -1 when the are of opposite directions. That is, the cosine similarity between two vectors scales with their relevance. So we can calculate the cosine similarity of two word embedding vectors to represent their relevance:
Y
Yu Yang 已提交
46

Z
zhangruiqing01 已提交
47 48
```
please input two words: big huge
C
choijulie 已提交
49
similarity: 0.899180685161
Z
zhangruiqing01 已提交
50 51 52 53 54

please input two words: from company
similarity: -0.0997506977351
```

55
The above results could be obtained by running `calculate_dis.py`, which loads the words in the dictionary and their corresponding trained word embeddings. For detailed instruction, see section [Model Application](https://github.com/PaddlePaddle/book/tree/develop/04.word2vec#model-application).
C
choijulie 已提交
56

Z
zhangruiqing01 已提交
57

C
choijulie 已提交
58
## Model Overview
Z
zhangruiqing01 已提交
59

C
choijulie 已提交
60
In this section, we will introduce three word embedding models: N-gram model, CBOW, and Skip-gram, which all output the frequency of each word given its immediate context.
Z
zhangruiqing01 已提交
61

62
For N-gram model, we will first introduce the concept of language model, and implement it using PaddlePaddle in section [Training](https://github.com/PaddlePaddle/book/tree/develop/04.word2vec#model-application).
Z
zhangruiqing01 已提交
63

M
Mimee 已提交
64
The latter two models, which became popular recently, are neural word embedding model developed by Tomas Mikolov at Google \[[3](#references)\]. Despite their apparent simplicity, these models train very well.
Z
zhangruiqing01 已提交
65

C
choijulie 已提交
66
### Language Model
Z
zhangruiqing01 已提交
67

C
choijulie 已提交
68 69 70 71 72 73
Before diving into word embedding models, we will first introduce the concept of **language model**. Language models build the joint probability function $P(w_1, ..., w_T)$ of a sentence, where $w_i$ is the i-th word in the sentence. The goal is to give higher probabilities to meaningful sentences, and lower probabilities to meaningless constructions.

In general, models that generate the probability of a sequence can be applied to many fields, like machine translation, speech recognition, information retrieval, part-of-speech tagging, and handwriting recognition. Take information retrieval, for example. If you were to search for "how long is a football bame" (where bame is a medical noun), the search engine would have asked if you had meant "how long is a football game" instead. This is because the probability of "how long is a football bame" is very low according to the language model; in addition, among all of the words easily confused with "bame", "game" would build the most probable sentence.

#### Target Probability
For language model's target probability $P(w_1, ..., w_T)$, if the words in the sentence were to be independent, the joint probability of the whole sentence would be the product of each word's probability:
Z
zhangruiqing01 已提交
74 75 76

$$P(w_1, ..., w_T) = \prod_{t=1}^TP(w_t)$$

C
choijulie 已提交
77
However, the frequency of words in a sentence typically relates to the words before them, so canonical language models are constructed using conditional probability in its target probability:
Z
zhangruiqing01 已提交
78 79 80

$$P(w_1, ..., w_T) = \prod_{t=1}^TP(w_t | w_1, ... , w_{t-1})$$

Z
zhangruiqing01 已提交
81

82
### N-gram neural model
Z
zhangruiqing01 已提交
83

C
choijulie 已提交
84
In computational linguistics, n-gram is an important method to represent text. An n-gram represents a contiguous sequence of n consecutive items given a text. Based on the desired application scenario, each item could be a letter, a syllable or a word. The N-gram model is also an important method in statistical language modeling. When training language models with n-grams, the first (n-1) words of an n-gram are used to predict the *n*th word.
Z
zhangruiqing01 已提交
85

M
Mimee 已提交
86
Yoshua Bengio and other scientists describe how to train a word embedding model using neural network in the famous paper of Neural Probabilistic Language Models \[[1](#references)\] published in 2003. The Neural Network Language Model (NNLM) described in the paper learns the language model and word embedding simultaneously through a linear transformation and a non-linear hidden connection. That is, after training on large amounts of corpus, the model learns the word embedding; then, it computes the probability of the whole sentence, using the embedding. This type of language model can overcome the **curse of dimensionality** i.e. model inaccuracy caused by the difference in dimensionality between training and testing data. Note that the term *neural network language model* is ill-defined, so we will not use the name NNLM but only refer to it as *N-gram neural model* in this section.
Z
zhangruiqing01 已提交
87

C
choijulie 已提交
88
We have previously described language model using conditional probability, where the probability of the *t*-th word in a sentence depends on all $t-1$ words before it. Furthermore, since words further prior have less impact on a word, and every word within an n-gram is only effected by its previous n-1 words, we have:
Z
zhangruiqing01 已提交
89

Z
zhangruiqing01 已提交
90
$$P(w_1, ..., w_T) = \prod_{t=n}^TP(w_t|w_{t-1}, w_{t-2}, ..., w_{t-n+1})$$
Z
zhangruiqing01 已提交
91

C
choijulie 已提交
92
Given some real corpus in which all sentences are meaningful, the n-gram model should maximize the following objective function:
Z
zhangruiqing01 已提交
93

Z
zhangruiqing01 已提交
94
$$\frac{1}{T}\sum_t f(w_t, w_{t-1}, ..., w_{t-n+1};\theta) + R(\theta)$$
Z
zhangruiqing01 已提交
95

96
where $f(w_t, w_{t-1}, ..., w_{t-n+1})$ represents the conditional logarithmic probability of the current word $w_t$ given its previous $n-1$ words, and $R(\theta)$ represents parameter regularization term.
Y
Yu Yang 已提交
97

98
<p align="center">
C
choijulie 已提交
99 100
       <img src="image/nnlm_en.png" width=500><br/>
       Figure 2. N-gram neural network model
Z
zhangruiqing01 已提交
101 102
</p>

103

C
choijulie 已提交
104 105 106
Figure 2 shows the N-gram neural network model. From the bottom up, the model has the following components:

 - For each sample, the model gets input $w_{t-n+1},...w_{t-1}$, and outputs the probability that the t-th word is one of `|V|` in the dictionary.
107

C
choijulie 已提交
108 109 110
 Every input word $w_{t-n+1},...w_{t-1}$ first gets transformed into word embedding $C(w_{t-n+1}),...C(w_{t-1})$ through a transformation matrix.

 - All the word embeddings concatenate into a single vector, which is mapped (nonlinearly) into the $t$-th word hidden representation:
111

112
    $$g=Utanh(\theta^Tx + b_1) + Wx + b_2$$
113

C
choijulie 已提交
114
   where $x$ is the large vector concatenated from all the word embeddings representing the context; $\theta$, $U$, $b_1$, $b_2$ and $W$ are parameters connecting word embedding layers to the hidden layers. $g$ represents the unnormalized probability of the output word, $g_i$ represents the unnormalized probability of the output word being the i-th word in the dictionary.
Z
zhangruiqing01 已提交
115

C
choijulie 已提交
116
 - Based on the definition of softmax, using normalized $g_i$, the probability that the output word is $w_t$ is represented as:
117

Z
zhangruiqing01 已提交
118
  $$P(w_t | w_1, ..., w_{t-n+1}) = \frac{e^{g_{w_t}}}{\sum_i^{|V|} e^{g_i}}$$
Z
zhangruiqing01 已提交
119

C
choijulie 已提交
120
 - The cost of the entire network is a multi-class cross-entropy and can be described by the following loss function
Z
zhangruiqing01 已提交
121

122
   $$J(\theta) = -\sum_{i=1}^N\sum_{c=1}^{|V|}y_k^{i}log(softmax(g_k^i))$$
Z
zhangruiqing01 已提交
123

C
choijulie 已提交
124
   where $y_k^i$ represents the true label for the $k$-th class in the $i$-th sample ($0$ or $1$), $softmax(g_k^i)$ represents the softmax probability for the $k$-th class in the $i$-th sample.
125 126

### Continuous Bag-of-Words model(CBOW)
Z
zhangruiqing01 已提交
127

C
choijulie 已提交
128
CBOW model predicts the current word based on the N words both before and after it. When $N=2$, the model is as the figure below:
Y
Yu Yang 已提交
129

130
<p align="center">
C
choijulie 已提交
131 132
    <img src="image/cbow_en.png" width=250><br/>
    Figure 3. CBOW model
Z
zhangruiqing01 已提交
133
</p>
Z
zhangruiqing01 已提交
134

C
choijulie 已提交
135
Specifically, by ignoring the order of words in the sequence, CBOW uses the average value of the word embedding of the context to predict the current word:
Z
zhangruiqing01 已提交
136

C
choijulie 已提交
137
$$\text{context} = \frac{x_{t-1} + x_{t-2} + x_{t+1} + x_{t+2}}{4}$$
Z
zhangruiqing01 已提交
138

C
choijulie 已提交
139
where $x_t$ is the word embedding of the t-th word, classification score vector is $z=U*\text{context}$, the final classification $y$ uses softmax and the loss function uses multi-class cross-entropy.
Z
zhangruiqing01 已提交
140

141
### Skip-gram model
Z
zhangruiqing01 已提交
142

C
choijulie 已提交
143
The advantages of CBOW is that it smooths over the word embeddings of the context and reduces noise, so it is very effective on small dataset. Skip-gram uses a word to predict its context and get multiple context for the given word, so it can be used in larger datasets.
Y
Yu Yang 已提交
144

145
<p align="center">
C
choijulie 已提交
146 147
    <img src="image/skipgram_en.png" width=250><br/>
    Figure 4. Skip-gram model
Z
zhangruiqing01 已提交
148 149
</p>

C
choijulie 已提交
150
As illustrated in the figure above, skip-gram model maps the word embedding of the given word onto $2n$ word embeddings (including $n$ words before and $n$ words after the given word), and then combine the classification loss of all those $2n$ words by softmax.
Z
zhangruiqing01 已提交
151

C
choijulie 已提交
152
## Dataset
Z
zhangruiqing01 已提交
153

M
Mimee 已提交
154
We will use Penn Treebank (PTB) (Tomas Mikolov's pre-processed version) dataset. PTB is a small dataset, used in Recurrent Neural Network Language Modeling Toolkit\[[2](#references)\]. Its statistics are as follows:
Z
zhangruiqing01 已提交
155 156 157

<p align="center">
<table>
158
    <tr>
C
choijulie 已提交
159 160 161
        <td>training set</td>
        <td>validation set</td>
        <td>test set</td>
162 163 164 165 166 167 168
    </tr>
    <tr>
        <td>ptb.train.txt</td>
        <td>ptb.valid.txt</td>
        <td>ptb.test.txt</td>
    </tr>
    <tr>
C
choijulie 已提交
169 170 171
        <td>42068 lines</td>
        <td>3370 lines</td>
        <td>3761 lines</td>
172
    </tr>
Z
zhangruiqing01 已提交
173 174 175
</table>
</p>

C
choijulie 已提交
176 177 178 179 180 181
### Python Dataset Module

We encapsulated the PTB Data Set in our Python module `paddle.dataset.imikolov`. This module can

1. download the dataset to `~/.cache/paddle/dataset/imikolov`, if not yet, and
2. [preprocesses](#preprocessing) the dataset.
182

C
choijulie 已提交
183
### Preprocessing
Z
zhangruiqing01 已提交
184

C
choijulie 已提交
185
We will be training a 5-gram model. Given five words in a window, we will predict the fifth word given the first four words.
Z
zhangruiqing01 已提交
186

C
choijulie 已提交
187
Beginning and end of a sentence have a special meaning, so we will add begin token `<s>` in the front of the sentence. And end token `<e>` in the end of the sentence. By moving the five word window in the sentence, data instances are generated.
Z
zhangruiqing01 已提交
188

C
choijulie 已提交
189
For example, the sentence "I have a dream that one day" generates five data instances:
Z
zhangruiqing01 已提交
190

191 192 193 194 195 196
```text
<s> I have a dream
I have a dream that
have a dream that one
a dream that one day
dream that one day <e>
Z
zhangruiqing01 已提交
197 198
```

C
choijulie 已提交
199
At last, each data instance will be converted into an integer sequence according it's words' index inside the dictionary.
Z
zhangruiqing01 已提交
200

C
choijulie 已提交
201 202 203
## Training

The neural network that we will be using is illustrated in the graph below:
Z
zhangruiqing01 已提交
204

205
<p align="center">
C
choijulie 已提交
206 207
    <img src="image/ngram.en.png" width=400><br/>
    Figure 5. N-gram neural network model in model configuration
Z
zhangruiqing01 已提交
208 209
</p>

C
choijulie 已提交
210 211
`word2vec/train.py` demonstrates training word2vec using PaddlePaddle:

212 213
### Datafeeder Configuration
Our program starts with importing necessary packages:
214

215
- Import packages.
216 217

```python
218 219 220
import paddle
import paddle.fluid as fluid
import numpy
221 222 223 224
from functools import partial
import math
import os
import sys
225 226
```

227
- Configure parameters and build word dictionary.
228

Y
Yu Yang 已提交
229
```python
230 231 232 233
EMBED_SIZE = 32  # word vector dimension
HIDDEN_SIZE = 256  # hidden layer dimension
N = 5  # train 5-gram
BATCH_SIZE = 32  # batch size
Z
zhangruiqing01 已提交
234

235 236
# can use CPU or GPU
use_cuda = os.getenv('WITH_GPU', '0') != '0'
237

Y
Yu Yang 已提交
238 239
word_dict = paddle.dataset.imikolov.build_dict()
dict_size = len(word_dict)
240 241
```

242
Unlike from the previous PaddlePaddle v2, in the new API (Fluid), we do not need to calculate word embedding ourselves. PaddlePaddle provides a built-in method `fluid.layers.embedding` and we can use it directly to build our N-gram neural network model.
243

244
- We define our N-gram neural network structure as below. This structure will be used both in `train` and in `infer`
245
```python
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283
def inference_program(is_sparse):
    first_word = fluid.layers.data(name='firstw', shape=[1], dtype='int64')
    second_word = fluid.layers.data(name='secondw', shape=[1], dtype='int64')
    third_word = fluid.layers.data(name='thirdw', shape=[1], dtype='int64')
    fourth_word = fluid.layers.data(name='fourthw', shape=[1], dtype='int64')

    embed_first = fluid.layers.embedding(
        input=first_word,
        size=[dict_size, EMBED_SIZE],
        dtype='float32',
        is_sparse=is_sparse,
        param_attr='shared_w')
    embed_second = fluid.layers.embedding(
        input=second_word,
        size=[dict_size, EMBED_SIZE],
        dtype='float32',
        is_sparse=is_sparse,
        param_attr='shared_w')
    embed_third = fluid.layers.embedding(
        input=third_word,
        size=[dict_size, EMBED_SIZE],
        dtype='float32',
        is_sparse=is_sparse,
        param_attr='shared_w')
    embed_fourth = fluid.layers.embedding(
        input=fourth_word,
        size=[dict_size, EMBED_SIZE],
        dtype='float32',
        is_sparse=is_sparse,
        param_attr='shared_w')

    concat_embed = fluid.layers.concat(
        input=[embed_first, embed_second, embed_third, embed_fourth], axis=1)
    hidden1 = fluid.layers.fc(input=concat_embed,
                              size=HIDDEN_SIZE,
                              act='sigmoid')
    predict_word = fluid.layers.fc(input=hidden1, size=dict_size, act='softmax')
    return predict_word
284 285
```

286
- As we already defined the N-gram neural network structure in the above, we can use it in our `train` method.
287 288

```python
289 290 291 292 293 294 295 296 297
def train_program(is_sparse):
    # The declaration of 'next_word' must be after the invoking of inference_program,
    # or the data input order of train program would be [next_word, firstw, secondw,
    # thirdw, fourthw], which is not correct.
    predict_word = inference_program(is_sparse)
    next_word = fluid.layers.data(name='nextw', shape=[1], dtype='int64')
    cost = fluid.layers.cross_entropy(input=predict_word, label=next_word)
    avg_cost = fluid.layers.mean(cost)
    return avg_cost
298
```
299

300
- Now we will begin the training process. It is relatively simple compared to the previous version.  `paddle.dataset.imikolov.train()` and `paddle.dataset.imikolov.test()` are our training and test set. Both of the functions will return a **reader**: In PaddlePaddle, reader is a python function which returns a Python iterator which output a single data instance at a time.
301

C
choijulie 已提交
302
`paddle.batch` takes reader as input, outputs a **batched reader**: In PaddlePaddle, a reader outputs a single data instance at a time but batched reader outputs a minibatch of data instances.
303

304 305
`event_handler` can be passed into `trainer.train` so that we can do some tasks after each step or epoch. These tasks include recording current metrics or terminate current training process.

306
```python
307 308 309 310 311 312
def optimizer_func():
    return fluid.optimizer.AdagradOptimizer(
        learning_rate=3e-3,
        regularization=fluid.regularizer.L2DecayRegularizer(8e-4))


313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329
def train(use_cuda, train_program, params_dirname):
    train_reader = paddle.batch(
        paddle.dataset.imikolov.train(word_dict, N), BATCH_SIZE)
    test_reader = paddle.batch(
        paddle.dataset.imikolov.test(word_dict, N), BATCH_SIZE)

    place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()

    def event_handler(event):
        if isinstance(event, fluid.EndStepEvent):
            outs = trainer.test(
                reader=test_reader,
                feed_order=['firstw', 'secondw', 'thirdw', 'fourthw', 'nextw'])
            avg_cost = outs[0]

            # We output cost every 10 steps.
            if event.step % 10 == 0:
330
                print "Step %d: Average Cost %f" % (event.step, avg_cost)
331

332 333 334
            # If average cost is lower than 5.8, we consider the model good enough to stop.
            # Note 5.8 is a relatively high value. In order to get a better model, one should
            # aim for avg_cost lower than 3.5. But the training could take longer time.
335
            if avg_cost < 5.8:
336 337 338 339 340 341 342 343 344 345 346 347
                trainer.save_params(params_dirname)
                trainer.stop()

            if math.isnan(avg_cost):
                sys.exit("got NaN loss, training failed.")

    trainer = fluid.Trainer(
        train_func=train_program,
        # Note here we need to choose more sophisticated optimizer
        # such as AdaGrad with a decay rate. The normal SGD converges
        # very slowly.
        # optimizer=fluid.optimizer.SGD(learning_rate=0.001),
348
        optimizer_func=optimizer_func,
349 350 351 352 353 354 355
        place=place)

    trainer.train(
        reader=train_reader,
        num_epochs=1,
        event_handler=event_handler,
        feed_order=['firstw', 'secondw', 'thirdw', 'fourthw', 'nextw'])
Z
zhangruiqing01 已提交
356 357
```

C
choijulie 已提交
358
`trainer.train` will start training, the output of `event_handler` will be similar to following:
359
```text
360 361 362
Step 0: Average Cost 7.337213
Step 10: Average Cost 6.136128
Step 20: Average Cost 5.766995
363 364
...
```
Y
Yu Yang 已提交
365

Z
zhangruiqing01 已提交
366

C
choijulie 已提交
367
## Model Application
368

369
After the model is trained, we can load the saved model parameters and do some inference.
Z
zhangruiqing01 已提交
370

371
### Predicting the next word
Z
zhangruiqing01 已提交
372

373
We can use our trained model to predict the next word given its previous N-gram. For example
Z
zhangruiqing01 已提交
374 375


376 377 378 379 380 381 382 383 384 385 386 387 388
```python
def infer(use_cuda, inference_program, params_dirname=None):
    place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
    inferencer = fluid.Inferencer(
        infer_func=inference_program, param_path=params_dirname, place=place)

    # Setup inputs by creating 4 LoDTensors representing 4 words. Here each word
    # is simply an index to look up for the corresponding word vector and hence
    # the shape of word (base_shape) should be [1]. The length-based level of
    # detail (lod) info of each LoDtensor should be [[1]] meaning there is only
    # one lod_level and there is only one sequence of one word on this level.
    # Note that lod info should be a list of lists.

389 390 391 392 393 394 395 396 397 398
    data1 = [[211]]  # 'among'
    data2 = [[6]]    # 'a'
    data3 = [[96]]   # 'group'
    data4 = [[4]]    # 'of'
    lod = [[1]]

    first_word  = fluid.create_lod_tensor(data1, lod, place)
    second_word = fluid.create_lod_tensor(data2, lod, place)
    third_word  = fluid.create_lod_tensor(data3, lod, place)
    fourth_word = fluid.create_lod_tensor(data4, lod, place)
399 400 401 402 403 404 405 406 407 408 409 410 411

    result = inferencer.infer(
        {
            'firstw': first_word,
            'secondw': second_word,
            'thirdw': third_word,
            'fourthw': fourth_word
        },
        return_numpy=False)

    print(numpy.array(result[0]))
    most_possible_word_index = numpy.argmax(result[0])
    print(most_possible_word_index)
412 413 414 415
    print([
        key for key, value in word_dict.iteritems()
        if value == most_possible_word_index
    ][0])
Z
zhangruiqing01 已提交
416 417
```

418
When we spent 3 mins in training, the output is like below, which means the next word for `among a group of` is `a`. If we train the model with a longer time, it will give a meaningful prediction as `workers`.
419

420
```text
421 422 423
[[0.00106646 0.0007907  0.00072041 ... 0.00049024 0.00041355 0.00084464]]
6
a
424
```
Y
Yu Yang 已提交
425

426
The main entrance of the program is fairly simple:
Z
zhangruiqing01 已提交
427

Y
Yu Yang 已提交
428
```python
429 430 431
def main(use_cuda, is_sparse):
    if use_cuda and not fluid.core.is_compiled_with_cuda():
        return
Z
zhangruiqing01 已提交
432

433
    params_dirname = "word2vec.inference.model"
Z
zhangruiqing01 已提交
434

435 436 437 438
    train(
        use_cuda=use_cuda,
        train_program=partial(train_program, is_sparse),
        params_dirname=params_dirname)
Z
zhangruiqing01 已提交
439

440 441 442 443
    infer(
        use_cuda=use_cuda,
        inference_program=partial(inference_program, is_sparse),
        params_dirname=params_dirname)
Y
Yu Yang 已提交
444

Z
zhangruiqing01 已提交
445

446
main(use_cuda=use_cuda, is_sparse=True)
447
```
448

C
choijulie 已提交
449 450
## Conclusion

451
This chapter introduces word embeddings, the relationship between language model and word embedding, and how to train neural networks to learn word embedding.
C
choijulie 已提交
452

453
In grammar analysis and semantic analysis, a previously trained word embedding can initialize models for better performance. We hope that readers can use word embedding models in their work after reading this chapter.
Z
zhangruiqing01 已提交
454 455


M
Mimee 已提交
456
## References
Z
zhangruiqing01 已提交
457
1. Bengio Y, Ducharme R, Vincent P, et al. [A neural probabilistic language model](http://www.jmlr.org/papers/volume3/bengio03a/bengio03a.pdf)[J]. journal of machine learning research, 2003, 3(Feb): 1137-1155.
Z
zhangruiqing01 已提交
458 459 460 461
2. Mikolov T, Kombrink S, Deoras A, et al. [Rnnlm-recurrent neural network language modeling toolkit](http://www.fit.vutbr.cz/~imikolov/rnnlm/rnnlm-demo.pdf)[C]//Proc. of the 2011 ASRU Workshop. 2011: 196-201.
3. Mikolov T, Chen K, Corrado G, et al. [Efficient estimation of word representations in vector space](https://arxiv.org/pdf/1301.3781.pdf)[J]. arXiv preprint arXiv:1301.3781, 2013.
4. Maaten L, Hinton G. [Visualizing data using t-SNE](https://lvdmaaten.github.io/publications/papers/JMLR_2008.pdf)[J]. Journal of Machine Learning Research, 2008, 9(Nov): 2579-2605.
5. https://en.wikipedia.org/wiki/Singular_value_decomposition
L
Luo Tao 已提交
462 463

<br/>
L
Luo Tao 已提交
464
This tutorial is contributed by <a xmlns:cc="http://creativecommons.org/ns#" href="http://book.paddlepaddle.org" property="cc:attributionName" rel="cc:attributionURL">PaddlePaddle</a>, and licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike 4.0 International License</a>.