-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathsem_SC_MSA.py
292 lines (265 loc) · 17.4 KB
/
sem_SC_MSA.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
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
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
284
285
286
287
288
289
290
291
import sys
sys.path.append('..')
import tensorflow as tf
import pickle
import random
import numpy as np
import math
import os
import utils
if __name__ == '__main__':
# ################ Prepare Data ###################
# basic config
dim = 200 # dimension of embedding
lr_decay_rate = 0.99 # learning rate decay rate
batch_size = 1 # batch size, set to 1 because we use SGD
learning_rate = 0.2 # initial learning rate 0.2
total_epoch = 40 # total training epoches 40
k = 100 # parameter k for weighted_cross_entropy_with_logits
hownet_filename = 'dataset/hownet.txt'
comp_filename = 'dataset/all.bin'
train_filename = 'dataset/train.bin'
test_filename = 'dataset/test.bin'
dev_filename = 'dataset/dev.bin'
embedding_filename = 'dataset/word_embedding.txt'
sem_embed_filename = 'dataset/sememe_vector.txt'
logdir_name = 'sememe_prediction/SCMSA'
# load hownet，并把hownet.comp分成test_set和train_set
hownet = utils.Hownet(hownet_file=hownet_filename, comp_file=comp_filename)
hownet.build_hownet()
hownet.token2id()
hownet.load_split_dataset(train_filename=train_filename, test_filename=test_filename, dev_filename=dev_filename)
word_embedding_np, hownet = utils.load_word_embedding(embedding_filename, hownet, scale=True) # load word embedding
sememe_embedding_np = utils.load_sememe_embedding(sem_embed_filename, hownet, scale=True) # load sememe embedding
train_num = len(hownet.comp_train)
pos_dict, word_remove = utils.load_hownet_pos()
hownet, cls_dict = utils.divide_data_with_pos(pos_dict, hownet)
print("number of dataset in training set:{}".format(len(hownet.comp_train)))
print("number of dataset in test set:{}".format(len(hownet.comp_test)))
print("number of dataset in dev set:{}".format(len(hownet.comp_dev)))
if not os.path.exists(logdir_name):
os.makedirs(logdir_name)
os.makedirs(os.path.join(logdir_name, 'print_files'))
os.makedirs(os.path.join(logdir_name, 'model_file'))
os.makedirs(os.path.join(logdir_name, 'tensorboard_logs'))
os.makedirs(os.path.join(logdir_name, 'example_files'))
# ################ Prepare Data ###################
# ################ Model and Run ###################
input_word_l = tf.placeholder(tf.int32, shape=[1], name='word_left')
input_word_r = tf.placeholder(tf.int32, shape=[1], name='word_right')
input_sememes_l = tf.placeholder(tf.int32, shape=[None], name='sememes_left')
input_sememes_r = tf.placeholder(tf.int32, shape=[None], name='sememes_right')
answer_sememes = tf.placeholder(tf.int32, shape=[1, None], name='sememes_answer')
labels = tf.placeholder(tf.float32, shape=[1, hownet.sem_num], name='labels')
ones = tf.ones([1, hownet.sem_num], dtype=tf.float32)
tf_dim = tf.constant(math.log(dim), dtype=tf.float32, name='tf_dim')
word_embedding = tf.Variable(tf.constant(0.0, shape=[word_embedding_np.shape[0], word_embedding_np.shape[1]]),trainable=False,name='word_embed')
embedding_placeholder = tf.placeholder(tf.float32, [word_embedding_np.shape[0], word_embedding_np.shape[1]])
embedding_init = word_embedding.assign(embedding_placeholder)
sememe_embedding = tf.Variable(tf.constant(0.0, shape=[hownet.sem_num, dim]), trainable=True, name='Sememe_embeddings')
sememe_placeholder = tf.placeholder(tf.float32, [hownet.sem_num, dim])
sememe_init = sememe_embedding.assign(sememe_placeholder)
W_a = tf.Variable(tf.truncated_normal([dim, dim], stddev=0.5), tf.float32, name='W_a')
b_a = tf.Variable(tf.zeros([1, dim]), tf.float32, name='b_a')
W_c = tf.Variable(tf.truncated_normal([2 * dim, dim], stddev=1.0), tf.float32, name='W_c')
b_c = tf.Variable(tf.zeros([1, dim]), tf.float32, name='b_c')
global_step = tf.Variable(0, trainable=False)
with tf.name_scope('word_embedding'):
embed_word_r = tf.nn.embedding_lookup(word_embedding, input_word_r)
embed_word_l = tf.nn.embedding_lookup(word_embedding, input_word_l)
with tf.name_scope('sememe_embedding'):
embed_sememe_l = tf.nn.embedding_lookup(sememe_embedding, input_sememes_l)
embed_sememe_r = tf.nn.embedding_lookup(sememe_embedding, input_sememes_r)
embed_sememe_r = utils.norm(embed_sememe_r)
embed_sememe_l = utils.norm(embed_sememe_l)
with tf.name_scope('attention_left'):
embed_word_align_r = tf.nn.tanh(tf.matmul(embed_word_r, W_a)+b_a, name='embed_align_r') # 1 * dim
att_l = tf.nn.softmax(tf.matmul(embed_sememe_l, tf.transpose(embed_word_align_r)), axis=0)
embed_sememe_l = att_l * embed_sememe_l
embed_aggre_word_l_pure = tf.reduce_sum(embed_sememe_l, axis=0, keepdims=True, name='embed_aggre_word_l_pure')
with tf.name_scope('attention_right'):
embed_word_align_l = tf.nn.tanh(tf.matmul(embed_word_l, W_a)+b_a, name='embed_align_l')
att_r = tf.nn.softmax(tf.matmul(embed_sememe_r, tf.transpose(embed_word_align_l)), axis=0)
embed_sememe_r = att_r * embed_sememe_r
embed_aggre_word_r_pure = tf.reduce_sum(embed_sememe_r, axis=0, keepdims=True, name='embed_aggre_word_r_pure')
with tf.name_scope('phrase_embedding'):
embed_word_whole = embed_word_r + embed_word_l
embed_sememe_whole = embed_aggre_word_r_pure + embed_aggre_word_l_pure
phrase_vec = tf.nn.tanh(tf.matmul(tf.concat([embed_word_whole, embed_sememe_whole], 1), W_c)+b_c, name="phrase_vec")
with tf.name_scope('output_layer'):
y_hat = tf.matmul(phrase_vec, tf.transpose(sememe_embedding))
with tf.name_scope('cross_entropy_loss'):
cross_entropy = tf.nn.weighted_cross_entropy_with_logits(logits=y_hat, targets=labels, pos_weight=k, name='cross_entropy')
losses_pure = tf.reduce_mean(cross_entropy, name='cross_entropy_mean')
tf.summary.scalar("loss pure", losses_pure)
tf.add_to_collection("losses", losses_pure)
cross_entropy_mean = tf.add_n(tf.get_collection("losses")) # 只是变量名叫cross_entropy_mean, 实际上是loss_all
tf.summary.scalar('loss_all', cross_entropy_mean)
rank = tf.nn.top_k(tf.nn.sigmoid(y_hat), k=hownet.sem_num, name='rank')
lr = tf.train.exponential_decay(learning_rate, global_step=global_step, decay_steps=train_num / batch_size,decay_rate=lr_decay_rate, staircase=True)
tf.summary.scalar("learning_rate", lr)
opt = tf.train.GradientDescentOptimizer(learning_rate=lr)
grads_vars = opt.compute_gradients(cross_entropy_mean)
capped_grads_vars = [(tf.clip_by_value(grad, -5., 5.), var) for grad, var in grads_vars]
train_one_example = opt.apply_gradients(capped_grads_vars)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
sess.run(embedding_init, feed_dict={embedding_placeholder: word_embedding_np}) # initialize word embedding
sess.run(sememe_init, feed_dict={sememe_placeholder: sememe_embedding_np}) # initialize sememe embedding
merged = tf.summary.merge_all()
train_writer = tf.summary.FileWriter(logdir_name + '/tensorboard_logs', sess.graph) # writer of tensorboard
print_writer_filename = logdir_name + '/print_files/print.txt' # saver for printing
saver = tf.train.Saver(max_to_keep=3) # saver for saving model
# 这四个参数用来判断是否终止训练
# these 4 params. are used for deciding whether to stop training
last_map = 0
last_last_map = 10
now_map = 100
jump2test = False
random.shuffle(hownet.comp_train)
for epoch in range(total_epoch):
example_writer_filename = logdir_name + '/example_files/epoch' + str(epoch + 1) + '.txt' # file for writing examples
# train process
maps_train = []
loss_train = 0
for current_num, train_tup in enumerate(hownet.comp_train):
total_num = epoch * train_num + current_num
batch_dict = utils.generate_one_example4sememe_prediction(hownet, train_tup)
_, summary_i, step_num, loss_i, rank_res = sess.run(
[train_one_example, merged, global_step, cross_entropy_mean, rank],
feed_dict={input_word_l: [batch_dict['wl']],
input_word_r: [batch_dict['wr']],
input_sememes_l: batch_dict['sl'],
input_sememes_r: batch_dict['sr'],
labels: batch_dict['lb'],
answer_sememes: batch_dict['al']})
map_score = utils.cal_map_one(batch_dict['al'], rank_res[1])
maps_train.append(map_score)
loss_train += loss_i
if current_num % 100 == 0:
sys.stdout.flush()
sys.stdout.write('\rTraining num: ' + str(current_num) + ' of ' + str(train_num) + '.Epoch:' + str(epoch + 1))
loss_train = 0
train_writer.add_summary(summary=summary_i, global_step=step_num)
saver.save(sess, logdir_name + '/model_file/model_ckpt', global_step=epoch + 1)
# dev set test
maps_dev = []
loss_dev = 0
for current_num, dev_tup in enumerate(hownet.comp_dev):
batch_dev = utils.generate_one_example4sememe_prediction(hownet, dev_tup)
rank_res, loss_i = sess.run([rank, cross_entropy_mean],
feed_dict={input_word_l: [batch_dev['wl']],
input_word_r: [batch_dev['wr']],
input_sememes_l: batch_dev['sl'],
input_sememes_r: batch_dev['sr'],
labels: batch_dev['lb'],
answer_sememes: batch_dev['al']})
map_score = utils.cal_map_one(batch_dev['al'], rank_res[1])
maps_dev.append(map_score)
loss_dev += loss_i
print('Loss(dev )in epoch %d : %f' % (epoch + 1, loss_dev / len(hownet.comp_dev)))
print('MAP(dev) in epoch %d : %f' % (epoch + 1, sum(maps_dev) / float(len(hownet.comp_dev))))
print("*************DEV END*************\n")
# write log file
with open(print_writer_filename, 'a', encoding='utf-8') as fp:
fp.write('\nLoss(dev)in epoch %d:\t%f' % (epoch + 1, loss_dev / len(hownet.comp_dev)))
fp.write('\nMAP(dev ) in epoch %d:\t%f' % (epoch + 1, sum(maps_dev) / float(len(hownet.comp_dev))))
fp.write("**************DEV END*************\n")
# 判断终止条件：至少训练了20个epoch后，若在开发集上，连续两次map值上升，则终止；
# Deciding the stop condition: After training at least 20 epochs,
# if the MAP value rises twice in the development set, it stops;
if epoch+1 == 20:
last_map = sum(maps_dev) / float(len(hownet.comp_dev))
if epoch+1 == 21:
last_last_map = last_map
last_map = sum(maps_dev) / float(len(hownet.comp_dev))
elif epoch+1 >= 22:
now_map = sum(maps_dev) / float(len(hownet.comp_dev))
if now_map <= last_map <= last_last_map:
jump2test = True
else:
last_last_map = last_map
last_map = now_map
if epoch+1 >= 40:
jump2test = True
if jump2test:
model_file = os.path.join(logdir_name, 'model_file')
if not os.path.exists(model_file):
print("WARNING: path doesn't exist!")
sys.exit(0)
files = os.listdir(model_file)
third_last = 'model_ckpt-99'
for _model in files:
if _model != 'checkpoint':
_model = _model[:13]
if _model < third_last: # 加载倒数第二次map最大的那个文件
third_last = _model
epoch = int(third_last[11:13]) - 1
meta_file = os.path.join(model_file, third_last + '.meta')
data_file = os.path.join(model_file, third_last + '.data-00000-of-00001')
phrase_vec_file = os.path.join(logdir_name, 'example_files', 'phrase_vector.txt')
third_last = os.path.join(model_file, third_last)
saver.restore(sess, third_last)
# train set test
maps_test = []
loss_test = 0
for current_num, train_tup in enumerate(hownet.comp_train):
batch_train = utils.generate_one_example4sememe_prediction(hownet, train_tup)
rank_res, loss_i, att_left, att_right = sess.run([rank, cross_entropy_mean, att_l, att_r],
feed_dict={input_word_l: [batch_train['wl']],
input_word_r: [batch_train['wr']],
input_sememes_l: batch_train['sl'],
input_sememes_r: batch_train['sr'],
labels: batch_train['lb'],
answer_sememes: batch_train['al']})
map_score = utils.cal_map_one(batch_train['al'], rank_res[1])
maps_test.append(map_score)
loss_test += loss_i
print("************TRAIN START*************")
print('Loss(train )in epoch %d : %f' % (epoch + 1, loss_test / len(hownet.comp_train)))
print('MAP(train) in epoch %d : %f' % (epoch + 1, sum(maps_test) / float(len(hownet.comp_train))))
print("************TRAIN END***************\n")
# write log file
with open(print_writer_filename, 'a', encoding='utf-8') as fp:
fp.write("\n************TRAIN START*************\n")
fp.write('Loss(train)in epoch %d : %f' % (epoch+1, loss_test/len(hownet.comp_train)))
fp.write('\nMAP(train) in epoch %d : %f'%(epoch+1,sum(maps_test)/float(len(hownet.comp_train))))
fp.write("************TRAIN END***************\n")
# test set test
maps_test = []
loss_test = 0
for current_num, test_tup in enumerate(hownet.comp_test):
batch_test = utils.generate_one_example4sememe_prediction(hownet, test_tup)
rank_res, loss_i, att_left, att_right = sess.run([rank, cross_entropy_mean, att_l, att_r],
feed_dict={input_word_l: [batch_test['wl']],
input_word_r: [batch_test['wr']],
input_sememes_l: batch_test['sl'],
input_sememes_r: batch_test['sr'],
labels: batch_test['lb'],
answer_sememes: batch_test['al']})
map_score = utils.cal_map_one(batch_test['al'], rank_res[1])
maps_test.append(map_score)
_, test_predict = utils.hamming_loss(batch_test['al'], rank_res[1], get_answer=True, predict_num=hownet.sem_num)
loss_test += loss_i
if len(test_predict) != 0:
test_predict_str = utils.predictlabel2char(hownet.id2sememe, test_predict)
with open(example_writer_filename, 'a', encoding='utf-8') as ex:
ex.write(test_tup[4] + '\n\t')
for s in test_predict_str['truth']:
ex.write(s + ' ')
ex.write('\n\t')
for s in test_predict_str['predict']:
ex.write(s + ' ')
ex.write('\n')
print("************TEST START*************")
print('Loss(test )in epoch %d : %f'%(epoch+1,loss_test/len(hownet.comp_test)))
print('MAP(test) in epoch %d : %f'%(epoch+1,sum(maps_test)/float(len(hownet.comp_test))))
print("************TEST END***************\n")
# write log file
with open(print_writer_filename, 'a', encoding='utf-8') as fp:
fp.write("\n************TEST START*************\n")
fp.write('Loss(test )in epoch %d : %f'%(epoch+1,loss_test/len(hownet.comp_test)))
fp.write('\nMAP(test) in epoch %d : %f'%(epoch+1,sum(maps_test)/float(len(hownet.comp_test))))
fp.write("************TEST END***************\n")
break
train_writer.close()