-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathmain_story_clozed_task.py
199 lines (155 loc) · 8.35 KB
/
main_story_clozed_task.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
# -*- coding: utf-8 -*-
"""
Created on Wed May 23 11:34:47 2018
@author: mauro, dario
Description:
The target of this NLU project was to successfully accomplish the story
cloze task. This task was developed to measure advanced commonsense understanding within every-
day written stories. To solve the task, one has to determine the correct ending sentence out of two
available options, given the four initial context sentences of the story.
"""
import tensorflow as tf
import numpy as np
import os, sys
import json
from feature_extraction import perform_sentence_embedding, generate_XY, \
perform_PCA, createFeatures, test_train_split
from recurrent_neural_network import RNN_class
from feature_extraction import PCA_on_features
def main(baseline, rnn_settings, fileName):
"""
main method to train the RNN and test it on the story clozed task
"""
# set paths
pathToModel = os.path.dirname(os.path.realpath(sys.argv[0]))
pathToData = os.path.join(pathToModel, 'data')
pathToLog = os.path.join(pathToModel, 'log')
pathToGraph = os.path.join(pathToModel, 'graph')
# embed stories
save_valid_name='pre_calc_emb_LSTM'
# load embeddings
try:
valid_stories = np.load(os.path.join(pathToData, save_valid_name+'_valid.npy'))
valid_stories_2 = np.load(os.path.join(pathToData, save_valid_name+'_valid_2.npy'))
train_stories = np.load(os.path.join(pathToData, save_valid_name+'_train.npy'))
test_stories = np.load(os.path.join(pathToData, save_valid_name+'_test.npy'))
print('loaded pre calculated embeddings')
except:
print('not able to load precalculated embeddings')
print('caluculating embedings, this might take a while...')
train_stories, valid_stories, valid_stories_2, \
test_stories = perform_sentence_embedding(pathToData = pathToData,
embedding=True,
fileName = fileName)
np.save(os.path.join(pathToData, save_valid_name+'_valid'), valid_stories)
np.save(os.path.join(pathToData, save_valid_name + '_valid_2'), valid_stories_2)
np.save(os.path.join(pathToData, save_valid_name+'_train'), train_stories)
np.save(os.path.join(pathToData, save_valid_name+'_test'), test_stories)
if rnn_settings['pca']:
# perform PCA on the embedded sentences
train_stories = perform_PCA(train_stories, rnn_settings['embedding_size'])
valid_stories = perform_PCA(valid_stories, rnn_settings['embedding_size'])
X_sentence_embedding, y_sentence_embedding = generate_XY(train_stories)
X_train, X_test, y_train, y_test = test_train_split(X_sentence_embedding, y_sentence_embedding, 80000)
# create rnn graph
RNN = RNN_class(rnn_settings)
# build the graph of the RNN
RNN.build_graph(is_training = True)
number_of_paramters = RNN.get_num_parameters()
print('Number of Model Paramters:', number_of_paramters)
print('Number of Model Paramters in Millions:', number_of_paramters/10**6)
# Launch the graph
with tf.Session() as session:
if rnn_settings['Training_mode']:
# create rnn graph for the test case
RNN_validate = RNN_class(rnn_settings)
# build the graph of the RNN for the test case
RNN_validate.reuseVar=True
RNN_validate.build_graph(is_training = False)
saver = tf.train.Saver()
# Initialize the variables
session.run(tf.global_variables_initializer())
# train the model
writer_train = tf.summary.FileWriter(os.path.join(pathToLog, 'train'))
writer_train.add_graph(session.graph)
writer_validate = tf.summary.FileWriter(os.path.join(pathToLog, 'validate'))
writer_validate.add_graph(session.graph)
global_step = 0
# iterate over all epochs
for epoch_i in range(RNN.number_of_epochs):
# shuffle the training data
X_train, y_train = RNN.shuffleData(epoch_i, X_train, y_train)
print('#epoch: ', epoch_i)
# train the RNN
global_step = RNN.train(session, X_train, y_train, writer_train, global_step)
# validate the RNN
RNN_validate.validate(session, X_test, y_test, writer_validate, global_step)
# export the trained meta-graph
saver.save(session, os.path.join(pathToGraph, rnn_settings['save settings name'] + '.ckpt'))
with open(os.path.join(pathToGraph, rnn_settings['save settings name'] + '.json'), 'w') as fp:
json.dump(rnn_settings, fp)
else:
#predict the embedding of sentence
saver = tf.train.Saver()
saver.restore(session, os.path.join(pathToGraph, rnn_settings['save settings name'] + '.ckpt'))
print('model restored!')
if baseline:
X_true = valid_stories[:,4,:]
X_false = valid_stories[:,5,:]
X_valid_true = valid_stories_2[:,4,:]
X_valid_false = valid_stories_2[:,5,:]
X_test_sent_1 = test_stories[:,4,:]
X_test_sent_2 = test_stories[:,5,:]
else:
X_true, X_false = createFeatures(mode = 'valid',
stories = valid_stories,
session = session,
pathToData = pathToData,
pathToModel = pathToModel,
RNN = RNN,
fileName = fileName)
X_valid_true, X_valid_false = createFeatures(mode = 'valid_2',
stories = valid_stories_2,
session = session,
pathToData = pathToData,
pathToModel = pathToModel,
RNN = RNN,
fileName = fileName)
X_test_sent_1, X_test_sent_2 = createFeatures(mode = 'test',
stories = test_stories,
session = session,
pathToData = pathToData,
pathToModel = pathToModel,
RNN = RNN,
fileName = fileName)
RNN.classification(X_true, X_false, X_valid_true, X_valid_false)
RNN.createSubmissionFile(X_test_sent_1, X_test_sent_2)
if __name__ == "__main__":
# reset the built graph
tf.reset_default_graph()
tf.logging.set_verbosity(tf.logging.ERROR)
baseline = False
# define the rnn with LSTM cell
rnn_settings = {
'number_of_sentences' : 5,
'batch_size' : 16,
'embedding_size' : 512,
'lstm_size' : 128, #64,
'learning_rate' : 0.001, # 0.001
'number_of_epochs' : 80, #, 8,
'clip_gradient' : 10.0,
'num_layers': 3,
'dropout_rate': 0.3,
'decay_step': 4000,
'save settings name': 'LSTM3layer80epochLSTM128_Emb512batchsize16clip10_dropout0lr0.001Bidirecitional_try_overfit', # 'MSE_LSTM3layer7epochLSTM512_Emb512batchsize8clip10_dropout0lr0.001',
'Training_mode': True,
'pca': False,
'bidirectional_rnn' : True
}
# set the data file names
fileName = {'train': 'train_stories.csv',
'test': 'test_nlu18_utf-8.csv',
'valid': 'cloze_test_val__spring2016_cloze_test_ALL_val.csv',
'valid_2': 'cloze_test_spring2016-test.csv'}
# run main method
main(baseline, rnn_settings, fileName)