-
Notifications
You must be signed in to change notification settings - Fork 24
/
run_BERT.py
377 lines (335 loc) · 20.2 KB
/
run_BERT.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
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
# -*- coding: utf-8 -*-
'''
# @project : SDGCN
# @Time : 2019/4/26 11:56
# @Author : plzhao
# @Software: PyCharm
'''
import numpy as np
import tensorflow as tf
import os
import time
import datetime
import data_helpers
from sklearn import metrics
from models.gcn_bert import GCN_BERT
# Parameters
# ==================================================
# "Restaurants" or "Laptops"
use_data = "Laptops"
# "GCN_BERT"
use_model = "GCN_BERT"
datas = {"Restaurants_train": "data/data_res/bert_embedding/Restaurants_Train_bert.txt",
"Restaurants_test": "data/data_res/bert_embedding/Restaurants_Test_bert.txt",
"Restaurants_embedding": 'data/data_res/Restaurants_glove.42B.300d.txt',
"Laptops_train": "data/data_lap/bert_embedding/Laptops_Train_bert.txt",
"Laptops_test": "data/data_lap/bert_embedding/Laptops_Test_bert.txt",
"Laptops_embedding": 'data/data_lap/Laptops_glove.42B.300d.txt'}
#Train model
tf.flags.DEFINE_string("which_relation", 'global', "use which relation.") #'adjacent','global','rule'
tf.flags.DEFINE_string("which_model", use_model, "Model isused.")
# Data loading params
tf.flags.DEFINE_string("which_data", use_data, "Data is used.")
tf.flags.DEFINE_string("train_file", datas[use_data+"_train"], "Train data source.")
tf.flags.DEFINE_string("test_file", datas[use_data+"_test"], "Test data source.")
#word embedding
tf.flags.DEFINE_string('embedding_file_path', datas[use_data+"_embedding"], 'embedding file')
tf.flags.DEFINE_integer('word_embedding_dim', 768, 'dimension of word embedding')
# Model Hyperparameters
tf.flags.DEFINE_integer("num_filters", 128, "Number of filters per filter size (default: 128)")
tf.flags.DEFINE_float("learning_rate", 1e-3, "learning_rate (default: 1e-3)")
tf.flags.DEFINE_float("dropout_keep_prob", 0.5, "Dropout keep probability (default: 0.5)")
tf.flags.DEFINE_float("l2_reg_lambda", 0.01, "L2 regularization lambda (default: 0.0)")
# Training parameters
tf.flags.DEFINE_integer("batch_size", 32, "Batch Size (default: 32)")
tf.flags.DEFINE_integer("num_epochs", 80, "Number of training epochs ")
tf.flags.DEFINE_integer("evaluate_every", 5, "Evaluate model on dev set after this many steps")
tf.flags.DEFINE_integer("checkpoint_every", 5, "Save model after this many steps")
tf.flags.DEFINE_integer("num_checkpoints", 1, "Number of checkpoints to store")
# Misc Parameters
tf.flags.DEFINE_boolean("allow_soft_placement", True, "Allow device soft device placement")
tf.flags.DEFINE_boolean("log_device_placement", False, "Log placement of ops on devices")
FLAGS = tf.flags.FLAGS
# FLAGS._parse_flags()
# print("\nParameters:")
# for attr, value in sorted(FLAGS.__flags.items()):
# print("{}={}".format(attr.upper(), value))
# print("")
def preprocess():
'''
read from the text file.
:return: sen word id:[324,1413,1,41,43,0,0,0]
sen len:[5]
sen max len :[8]
sen label:[0,0,1]
target word id:[34,154,0,0]
target len: [2]
target max len: [4]
targets word id :[[34,154,0,0],
[34,14,12,56],
[0,0,0,0]]
targets num = 2
targets len: [2,4,0]
targets max num:[3]
targets_relation_self = [[1,0,0],
[0,1,0],
[0.0.0]]
targets_relation_cross = [[0,1,0],
[1,0,0],
[0.0.0]]
'''
# Data Preparation
# ==================================================
# Load data
print("Loading data...")
train_x_str,train_target_str, train_y = data_helpers.load_data_and_labels(FLAGS.train_file)
dev_x_str,dev_target_str, dev_y = data_helpers.load_data_and_labels(FLAGS.test_file)
test_x_str, test_target_str, test_y = data_helpers.load_data_and_labels(FLAGS.test_file)
#word embedding ---> x[324,1413,1,41,43,0,0,0] y[0,1]
#word_id_mapping,such as apple--->23 ,w2v 23---->[vector]
word_id_mapping, w2v = data_helpers.load_w2v(FLAGS.embedding_file_path, 300)
max_document_length = max([len(x.split(" ")) for x in (train_x_str + dev_x_str + test_x_str)])
max_target_length = max([len(x.split(" ")) for x in (train_target_str + dev_target_str + test_target_str)])
#The targets ---->[[[141,23,45],[23,45,1,2],[2]], ...]
#The number of targets ----> [3, ...]
train_targets_str,train_targets_num = data_helpers.load_targets(FLAGS.train_file)
dev_targets_str,dev_targets_num = data_helpers.load_targets(FLAGS.test_file)
test_targets_str, test_targets_num = data_helpers.load_targets(FLAGS.test_file)
max_target_num = max([len(x) for x in (train_targets_str + test_targets_str)])
# sentence ---> word_id
train_x, train_x_len = data_helpers.word2id(train_x_str,word_id_mapping,max_document_length)
dev_x, dev_x_len = data_helpers.word2id(dev_x_str,word_id_mapping,max_document_length)
test_x, test_x_len = data_helpers.word2id(test_x_str,word_id_mapping,max_document_length)
# target ---> word_id
train_target, train_target_len = data_helpers.word2id(train_target_str,word_id_mapping,max_target_length)
dev_target, dev_target_len = data_helpers.word2id( dev_target_str,word_id_mapping,max_target_length)
test_target, test_target_len = data_helpers.word2id(test_target_str,word_id_mapping,max_target_length)
# targets ---> word_id
train_targets, train_targets_len = data_helpers.word2id_2(train_targets_str,word_id_mapping,max_target_length,max_target_num)
dev_targets, dev_targets_len = data_helpers.word2id_2(dev_targets_str,word_id_mapping,max_target_length,max_target_num)
test_targets, test_targets_len = data_helpers.word2id_2(test_targets_str,word_id_mapping,max_target_length,max_target_num)
#which one targets in all targets
train_target_whichone = data_helpers.get__whichtarget(train_targets_num, max_target_num)
test_target_whichone = data_helpers.get__whichtarget(test_targets_num, max_target_num)
# target position
train_target_position = data_helpers.get_position(FLAGS.train_file,max_document_length)
test_target_position = data_helpers.get_position(FLAGS.test_file,max_document_length)
train_targets_position = data_helpers.get_position_2(train_target_position,train_targets_num,max_target_num)
test_targets_position = data_helpers.get_position_2(test_target_position,test_targets_num,max_target_num)
if use_data == 'Restaurants':
train_x = np.load("data/data_res/bert_embedding/Res_Train_Embedding.npy") #([3608,80,768])
train_target = np.load("data/data_res/bert_embedding/Res_Train_target_Embedding.npy") #([3608,23,768])
train_targets = np.load("data/data_res/bert_embedding/Res_Train_targets_Embedding.npy") #([3608,13,23,768])
test_x = np.load("data/data_res/bert_embedding/Res_Test_Embedding.npy") #([1120,80,768])
test_target = np.load("data/data_res/bert_embedding/Res_Test_target_Embedding.npy") #([1120,23,768])
test_targets = np.load("data/data_res/bert_embedding/Res_Test_targets_Embedding.npy") #([1120,13,23,768])
if use_data == 'Laptops':
train_x = np.load("data/data_lap/bert_embedding/Lap_Train_Embedding.npy") #([3608,80,768])
train_target = np.load("data/data_lap/bert_embedding/Lap_Train_target_Embedding.npy") #([3608,23,768])
train_targets = np.load("data/data_lap/bert_embedding/Lap_Train_targets_Embedding.npy") #([3608,13,23,768])
test_x = np.load("data/data_lap/bert_embedding/Lap_Test_Embedding.npy") #([1120,80,768])
test_target = np.load("data/data_lap/bert_embedding/Lap_Test_target_Embedding.npy") #([1120,23,768])
test_targets = np.load("data/data_lap/bert_embedding/Lap_Test_targets_Embedding.npy") #([1120,13,23,768])
#Relation Matrix
#use test_target to creat the relation
train_relation_self,train_relation_cross = data_helpers.get_relation(train_targets_num, max_target_num,FLAGS.which_relation)
test_relation_self, test_relation_cross = data_helpers.get_relation(test_targets_num, max_target_num,FLAGS.which_relation)
Train = {'x':train_x, # int32(3608, 79, 768) train sentences input embeddingID
'T':train_target, # int32(3608, 23, 768) train target input embeddingID
'Ts':train_targets, # int32(3608, 13, 23, 768) train targets input embeddingID
'x_len':train_x_len, # int32(3608,) train sentences input len
'T_len':train_target_len, # int32(3608,) train target len
'Ts_len': train_targets_len, # int32(3608, 13) train targets len
'T_W': train_target_whichone, # int32(3608, 13) the ith number of all the targets
'T_P':train_target_position, # float32(3608, 79)
'Ts_P': train_targets_position, # float32(3608,13, 79)
'R_Self': train_relation_self, # int32(3608, 13, 13)
'R_Cross': train_relation_cross, # int32(3608, 13, 13)
'y': train_y, # int32(3608, 3)
}
Test = { 'x':test_x,
'T':test_target,
'Ts':test_targets,
'x_len':test_x_len,
'T_len':test_target_len,
'Ts_len': test_targets_len,
'T_W': test_target_whichone,
'T_P': test_target_position,
'Ts_P': test_targets_position,
'R_Self': test_relation_self,
'R_Cross': test_relation_cross,
'y': test_y,
}
#
# batches = data_helpers.batch_iter(
# list(zip(x_train, y_train)), FLAGS.batch_size, FLAGS.num_epochs)
print("Vocabulary Size: {:d}".format(len(word_id_mapping)))
print("Train/Dev/test split: {:d}/{:d}/{:d}".format(len(train_y), len(dev_y), len(test_y)))
return Train,Test, w2v
def train(Train, Test, word_embedding):
# Training
# ==================================================
with tf.Graph().as_default():
session_conf = tf.ConfigProto(
allow_soft_placement=FLAGS.allow_soft_placement,
log_device_placement=FLAGS.log_device_placement)
sess = tf.Session(config=session_conf)
with sess.as_default():
model = eval(use_model)(
sequence_length=Train['x'].shape[1],
target_sequence_length = Train['T'].shape[1],
targets_num_max = Train['Ts'].shape[1],
num_classes=Train['y'].shape[1],
word_embedding_dim = FLAGS.word_embedding_dim,
l2_reg_lambda=FLAGS.l2_reg_lambda)
writer = tf.summary.FileWriter("logs/LSTM_GCN3", sess.graph)
vs = tf.trainable_variables()
print('There are %d train_able_variables in the Graph: ' % len(vs))
for v in vs:
print(v)
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer(FLAGS.learning_rate)
grads_and_vars = optimizer.compute_gradients(model.loss)
train_op = optimizer.apply_gradients(grads_and_vars, global_step=global_step)
# Keep track of gradient values and sparsity (optional)
grad_summaries = []
for g, v in grads_and_vars:
if g is not None:
grad_hist_summary = tf.summary.histogram("{}/grad/hist".format(v.name), g)
sparsity_summary = tf.summary.scalar("{}/grad/sparsity".format(v.name), tf.nn.zero_fraction(g))
grad_summaries.append(grad_hist_summary)
grad_summaries.append(sparsity_summary)
grad_summaries_merged = tf.summary.merge(grad_summaries)
# Output directory for models and summaries
timestamp = str(int(time.time()))
out_dir = os.path.abspath(os.path.join(os.path.curdir, "runs", use_data,use_model))
print("Writing to {}\n".format(out_dir))
# Summaries for loss and accuracy
loss_summary = tf.summary.scalar("loss", model.loss)
acc_summary = tf.summary.scalar("accuracy", model.accuracy)
# Train Summaries
train_summary_op = tf.summary.merge([loss_summary, acc_summary, grad_summaries_merged])
train_summary_dir = os.path.join(out_dir, "summaries", "train")
train_summary_writer = tf.summary.FileWriter(train_summary_dir, sess.graph)
# Test summaries
test_summary_op = tf.summary.merge([loss_summary, acc_summary])
test_summary_dir = os.path.join(out_dir, "summaries", "test")
test_summary_writer = tf.summary.FileWriter(test_summary_dir, sess.graph)
# Checkpoint directory. Tensorflow assumes this directory already exists so we need to create it
checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints"))
checkpoint_prefix = os.path.join(checkpoint_dir, "model")
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
# else:
# raise Exception('The checkpoint_dir already exists:',checkpoint_dir)
saver = tf.train.Saver(tf.global_variables(), max_to_keep=FLAGS.num_checkpoints)
# Write vocabulary
# vocab_processor.save(os.path.join(out_dir, "vocab"))
# Initialize all variables
sess.run(tf.global_variables_initializer())
def train_step(x_batch,T_batch,Ts_batch,x_len_batch,T_len_batch,Ts_len_batch,R_Self_batch,R_Cross_batxh,T_W_batch,T_P_batch,Ts_P_batch,y_batch):
"""
A single training step
"""
feed_dict = {
model.input_x: x_batch,
model.input_target:T_batch,
model.input_targets_all:Ts_batch,
model.sen_len:x_len_batch,
model.target_len:T_len_batch,
model.targets_all_len_a:Ts_len_batch,
model.relate_self:R_Self_batch,
model.relate_cross:R_Cross_batxh,
model.target_which:T_W_batch,
model.target_position: T_P_batch,
model.targets_all_position_a: Ts_P_batch,
model.input_y: y_batch,
model.dropout_keep_prob: FLAGS.dropout_keep_prob
}
_, step, summaries, loss, accuracy = sess.run(
[train_op, global_step, train_summary_op, model.loss, model.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
train_summary_writer.add_summary(summaries, step)
def test_step(x_batch,T_batch,Ts_batch,x_len_batch,T_len_batch,Ts_len_batch,R_Self_batch,R_Cross_batxh,T_W_batch,T_P_batch,Ts_P_batch,y_batch, summary = None,writer=None):
"""
Evaluates model on a dev set
"""
feed_dict = {
model.input_x: x_batch,
model.input_target:T_batch,
model.input_targets_all:Ts_batch,
model.sen_len:x_len_batch,
model.target_len:T_len_batch,
model.targets_all_len_a:Ts_len_batch,
model.relate_self:R_Self_batch,
model.relate_cross:R_Cross_batxh,
model.target_which: T_W_batch,
model.target_position: T_P_batch,
model.targets_all_position_a: Ts_P_batch,
model.input_y: y_batch,
model.dropout_keep_prob: 1.0
}
step, summaries, loss, accuracy, softmax,true_y,predictions = sess.run(
[global_step, summary, model.loss, model.accuracy, model.softmax,model.true_y, model.predictions],
feed_dict)
if writer:
writer.add_summary(summaries, step)
return step, loss, true_y, predictions
# Generate batches
batches = data_helpers.batch_iter(
list(zip(Train['x'],Train['T'],Train['Ts'],Train['x_len'], Train['T_len'], Train['Ts_len'],
Train['R_Self'],Train['R_Cross'],Train['T_W'],Train['T_P'],Train['Ts_P'],Train['y'])), FLAGS.batch_size, FLAGS.num_epochs)
# Training loop. For each batch...
train_acc, dev_acc, test_acc, train_all_softmax, test_all_softmax = [], [], [], [], []
max_test_acc = 0
max_test_F1_macro = 0
for batch in batches:
x_batch,T_batch,Ts_batch,x_len_batch,T_len_batch,Ts_len_batch,R_Self_batch,R_Cross_batxh,T_W_batch,T_P_batch,Ts_P_batch,y_batch = zip(*batch)
train_step(x_batch,T_batch,Ts_batch,x_len_batch,T_len_batch,Ts_len_batch,R_Self_batch,R_Cross_batxh,T_W_batch,T_P_batch,Ts_P_batch,y_batch)
current_step = tf.train.global_step(sess, global_step)
if current_step % FLAGS.evaluate_every == 0:
print('\nBy now ,the max test acc is: ', max_test_acc)
print(' the max F1 score is: ', max_test_F1_macro)
print("\nEvaluation Text:")
loss = 0
true_y = np.array([])
predictions = np.array([])
batches_test = data_helpers.batch_iter2(
list(zip(Test['x'], Test['T'], Test['Ts'], Test['x_len'], Test['T_len'], Test['Ts_len'],
Test['R_Self'], Test['R_Cross'], Test['T_W'], Test['T_P'], Test['Ts_P'], Test['y'])),256, 1, shuffle=False)
for batch_test in batches_test:
x_batch, T_batch, Ts_batch, x_len_batch, T_len_batch, Ts_len_batch, R_Self_batch, R_Cross_batxh, T_W_batch, T_P_batch, Ts_P_batch, y_batch = zip(*batch_test)
step_i, loss_i, true_y_i, predictions_i = test_step(x_batch, T_batch, Ts_batch, x_len_batch, T_len_batch, Ts_len_batch,
R_Self_batch,R_Cross_batxh, T_W_batch, T_P_batch, Ts_P_batch, y_batch, summary = test_summary_op, writer=test_summary_writer)
loss += loss_i
true_y = np.concatenate([true_y,true_y_i])
predictions = np.concatenate([predictions, predictions_i])
accuracy = metrics.accuracy_score(true_y, predictions)
test_F1_i = metrics.f1_score(true_y, predictions, average='macro')
time_str = datetime.datetime.now().isoformat()
print("{}: step {}, loss {:g}, acc {:g}, F1 {:g}".format(time_str, current_step, loss, accuracy, test_F1_i))
test_acc_i = accuracy
test_acc.append(test_acc_i)
print('----------------------------------------------------------')
print("")
if current_step % FLAGS.checkpoint_every == 0:
if test_acc_i>max_test_acc:
path = saver.save(sess, checkpoint_prefix, global_step=current_step)
print("Saved model checkpoint to {}\n".format(path))
print('->>>>>>>>>>>>>>>>>>>>>>>')
max_test_step = current_step
max_test_acc = test_acc_i
if test_F1_i > max_test_F1_macro:
max_test_F1_macro = test_F1_i
print('max_test_step: ', max_test_step)
print('max_test_acc: ', max_test_acc)
print('max_test_F1_macro: ', max_test_F1_macro)
return train_acc, dev_acc, max_test_acc,max_test_F1_macro,max_test_step, train_all_softmax, test_all_softmax
if __name__ == '__main__':
#文件处理
Train, Test, word_embedding = preprocess()
#模型训练
train_acc, dev_acc, max_test_acc,max_test_F1_macro,max_test_step, train_all_softmax, test_all_softmax = train(Train, Test, word_embedding)